Principles of Modern Radar - Volume 3 - PDFCOFFEE.COM (2024)

Principles of Modern Radar

Principles of Modern Radar Vol. III: Radar Applications

William L. Melvin Georgia Institute of Technology

James A. Scheer Georgia Institute of Technology

Edison, NJ scitechpub.com

Published by SciTech Publishing, an imprint of the IET. www.scitechpub.com www.theiet.org

Copyright ’ 2014 by SciTech Publishing, Edison, NJ. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning or otherwise, except as permitted under Sections 107 or 108 of the 1976 United States Copyright Act, without either the prior written permission of the Publisher, or authorization through payment of the appropriate per-copy fee to the Copyright Clearance Center, 222 Rosewood Drive, Danvers, MA 01923, (978) 750-8400, fax (978) 646-8600, or on the web at copyright.com. Requests to the Publisher for permission should be addressed to The Institution of Engineering and Technology, Michael Faraday House, Six Hills Way, Stevenage, Herts, SG1 2AY, United Kingdom. While the author and publisher believe that the information and guidance given in this work are correct, all parties must rely upon their own skill and judgement when making use of them. Neither the author nor publisher assumes any liability to anyone for any loss or damage caused by any error or omission in the work, whether such an error or omission is the result of negligence or any other cause. Any and all such liability is disclaimed. Editor: Dudley R. Kay Cover Design: Brent Beckley 10 9 8 7 6 5 4 3 2 1 ISBN 978-1-89112-154-8 (hardback) ISBN 978-1-61353-032-0 (PDF) Typeset in India by MPS Limited Printed in the USA by Sheridan Ltd Printed in the UK by CPI Group (UK) Ltd, Croydon

Contents

Preface xi Reviewer Acknowledgements xv Editors and Contributors xvii

1

Radar Applications 1

1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9

Introduction 1 Historical Perspective 2 Radar Measurements 5 Radar Frequencies 6 Radar Functions 8 U.S. Military Radar Nomenclature Topics in Radar Applications 10 Comments 14 References 15

2

Continuous Wave Radar

2.1 2.2 2.3 2.4 2.5 2.6

Introduction 17 Continuous Wave Radar 21 Frequency Modulated CW Radar 26 Other CW Radar Waveform Designs 63 FMCW Radar Applications 67 References 82

3

MMW Radar Characteristics and Applications 87

3.1 3.2 3.3 3.4 3.5 3.6 3.7

Introduction 87 The MMW Spectrum 88 Propagation at Higher Frequency 89 Antenna Beamwidth Considerations 93 MMW Performance Limitations 94 Typical Seeker or Smart Munition Configuration MMW Radar Applications 108

9

17

98

v

vi

Contents

3.8 MMW Future Trends 112 3.9 Further Reading 113 3.10 References 114

4

Fire-Control Radar

117

4.1 4.2 4.3 4.4 4.5 4.6 4.7

Introduction 117 Airborne Fire-Control Radar 123 Surface-Based Fire-Control Radar 160 Electronic Counter Countermeasures 170 The ‘‘AN’’ Equipment-Designation System References 173 Further Reading 173

5

Airborne Pulse-Doppler Radar

5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9 5.10 5.11 5.12 5.13 5.14

Introduction 175 Geometry 177 The Doppler Shift and Motivation for Doppler Processing Range and Doppler Distribution of Clutter 185 Contours of Constant Doppler and Range 196 Example Scenario 199 Pulse-Doppler Conceptual Approach 203 Ambiguities, Folded Clutter, and Blind Zones 216 Overview of PRF Regimes 226 High PRF Mode 228 Medium PRF Mode 235 Low PRF Mode 246 Summary 248 References 249

6

Multifunction Phased Array Radar Systems 251

6.1 6.2 6.3 6.4 6.5 6.6 6.7

Introduction 251 Operational Concepts and Military Utilities 254 MPARS Sizing and Performance Evaluation 257 ESA Overview 262 Radar Control and Resource Management 268 MPARS Technologies 276 MPARS Testing and Evaluation 280

172

175

181

Contents

6.8 Netcentric MPARS Applications 6.9 References 283 6.10 Further Reading 283

281

7

Ballistic Missile Defense Radar

285

7.1 7.2 7.3 7.4 7.5 7.6 7.7

Introduction 285 BMD Radar System Requirements 292 Radar Development for Ballistic Missile Defense BMD Radar Design 307 BMD Radar Performance Estimation 312 References 321 Further Reading 322

8

Ground-Based Early Warning Radar (GBEWR): Technology and Signal Processing Algorithms 323

8.1 8.2 8.3 8.4 8.5 8.6 8.7 8.8 8.9

Introduction 323 Phased Array Antenna 335 Transceiver 342 Waveforms and Signal Processing 348 Tracking 352 Electronic Counter-Countermeasures (ECCM) Capabilities Special Functions 359 Conclusions and Further Reading 376 References 377

9

Surface Moving Target Indication 383

9.1 9.2 9.3 9.4 9.5 9.6 9.7 9.8 9.9 9.10 9.11 9.12

Introduction 383 SMTI Radar Operation 390 Signal Models 393 SMTI Metrics 400 Antenna and Waveform Considerations Clutter-Mitigation Approaches 410 Detection Processing 418 Angle and Doppler Estimation 421 Other Considerations 424 Summary 426 Further Reading 427 References 427

405

298

357

vii

viii

Contents

10

Space-Based SAR for Remote Sensing 431

10.1 10.2 10.3 10.4 10.5 10.6 10.7 10.8 10.9

Introduction 431 Historical Perspective 438 Orbits 451 Design Considerations for the Spaceborne SAR 457 Special Modes and Capabilities 473 Design Example: Germany’s TerraSAR-X 482 Summary 493 References 494 Further Reading 498

11

Passive Bistatic Radar

11.1 11.2 11.3 11.4 11.5 11.6 11.7 11.8 11.9

Introduction 499 Bistatic Radar 505 Passive Bistatic Radar Waveforms The Signal Environment 519 Passive Bistatic Radar Techniques Examples of Systems 527 Conclusions 536 References 537 Further Reading 540

12

Air Traffic Control Radar

12.1 12.2 12.3 12.4 12.5 12.6 12.7 12.8

Introduction – The Task of Air Traffic Control (ATC) System Requirements/Mission 552 Design Issues 558 The Future of ATC Radar 582 Summary 585 Further Reading 585 Acknowledgments 585 References 585

13

Weather Radar

13.1 13.2 13.3 13.4

Introduction 591 Typical Weather-Radar Hardware 595 The Radar-Range Equation for Weather Radar Doppler Processing 603

499

509 524

543

591

598

543

Contents

13.5 13.6 13.7 13.8 13.9 13.10

Hydrological Measurements 609 Characteristics of Some Meteorological Phenomena Sun Echoes and Roost Rings 623 Advanced Processing and Systems 623 References 632 Further Reading 634

14

Foliage-Penetrating Radar

14.1 14.2 14.3 14.4 14.5 14.6 14.7 14.8 14.9 14.10

Introduction 635 History of Battlefield Surveillance 637 Foliage-Penetrating SAR Collection Systems FOPEN Clutter Characteristics 645 Image Formation 654 Radio Frequency Interference 665 Target Detection and Characterization 676 Summary 684 References 685 Further Reading 688

15

Ground-Penetrating Radar

15.1 15.2 15.3 15.4 15.5

Overview 691 Pulsed Ground-Penetrating Radar System Design 697 GPR System Implementation and Test Results 731 Conclusions 746 References 746

16

Police Radar

615

635

642

691

749

Introduction 749 The History of Technologies that Enabled Police Radar 750 Review of Homodyne Radar Principles 751 The First Police Radar 753 The Cosine Error Caused by Improper Operation 754 The Next-Generation S-band Radar 755 The Move to X-band – 10 GHz 758 A Second Method Used to Achieve the Ferro-Magnetic Circulator Function 763 16.9 Moving Radar with Improved Detection Range Capability 764 16.1 16.2 16.3 16.4 16.5 16.6 16.7 16.8

ix

x

Contents

16.10 16.11 16.12 16.13 16.14 16.15 16.16 16.17 Index

Moving-Mode Police Radar Operation 766 Alternative Phase-Locked Loop Signal-Processing Approach 770 The Move to K-band Frequencies 771 Police Radar Moves to the Ka-band and Utilizes Digital Signal Processing 772 Other Police Operating Modes Made Possible by DSP 774 Summary 777 References 777 Further Reading 778 779

Preface Principles of Modern Radar: Radar Applications is the third of the three-volume series of what was originally designed to be accomplished in one volume. As the final volume of the set, it finishes the original vision of a complete yet bounded reference for radar technology. This volume describes fifteen different system applications or class of applications in more detail than can be found in Volumes I or II. As different as the applications described, there is a difference in how these topics are treated by the authors. Whereas in Volumes I and II there is strict adherence to chapter format and level of detail, this volume has a wider dynamic range of technical depth. Some system applications lend themselves to a deeper level of technical description than others.

What This Book Addresses Certainly, there are many applications for which radar technology can be applied. Each chapter in Principles of Modern Radar: Radar Applications discusses a particular (selected) application or class of applications for the use of radar as a sensor. Not all applications for radar as a sensor are addressed in this volume, nor could they be. However, a varied selection of applications are included, providing a fairly broad cross section of surface-based and aerospace systems, defense-oriented as well as commercial technologies, and European as well as American systems. It was difficult to determine which system applications should be selected for this volume. Some areas of technology are so new that intellectual property rights restricted us from developing a complete picture of those applications. In other cases, classification issues were at play. Even considering these issues, there are many other radar applications that might have been covered, and a selection had to be made. We hope you are pleased with our choices.

Why This Book Was Written The original vision for PoMR was to provide the radar community with a single resource that described the latest radar technology, as driven largely by advancements in digital signal-processing (DSP) capability. Since DSP technology is maturing at such a fast pace, the ability to employ advanced techniques grows with it. The growth of these new techniques influences the development of advanced antenna techniques as well as subsystem radio-frequency and intermediate frequency hardware. The first two volumes in this series describe basic principles, some of which are true for legacy systems and some of which have experienced relatively recent use, as well as specific advanced techniques in the use of this technology. So, the first two volumes provide a complete picture of radar technology from the first principles to the advanced techniques in use today. With the publication of the first two volumes, it was natural to complete the original vision by preparing this volume describing selected modern radar applications. xi

xii

Preface

Who Should Read This Book Different from Volumes I and II, this volume is not intended as a textbook for the university environment. Rather, it was originally developed to be largely readable by the layperson, who might not necessarily have all the mathematical and scientific background to fully appreciate the material in the first two volumes. That stated, this volume is also intended to fill in some detail, reinforce or expand on fundamental technological issues described in the first two volumes, and round out understanding of system issues, at least for a selection of applications.

How the Chapters Are Structured The framework for each chapter was written roughly to answer the following questions: What are the system requirements? What are the particular radar issues associated with these requirements? How specifically are these features incorporated in the system? Examples of specific systems representing the class of applications discussed herein support the answers to these questions. Since different radar technology communities sometimes use different, or unique, symbols and abbreviations, many chapters have a separate table of abbreviations and symbols. It would be more difficult to read if all of the abbreviations and symbols were consolidated at the end of the book. Since this volume is not expected to be used as a university text, no student questions are included.

The History of the PoMR Series As discussed in the prefaces of Volumes I and II, the PoMR series was originally planned as one volume, entitled Principles of Modern Radar: Basic Principles, Advanced Techniques, and Radar Applications. The resulting number of chapters and sheer amount of the material suggested two volumes: the Basic Principles volume and the Advanced Techniques and Radar Applications volume. True to form, as Volume II emerged, it was separated into two volumes, resulting in the current set of three volumes. Volume I was written to provide a modern look at the fundamental technology and design issues related to radar technology in general. It provides an in-depth look at the modern signal-processing techniques available today, many that were not supported by the computing resources (signal- and data-processing technology) available even ten years ago. Volume II was prepared to demonstrate specific signal-processing techniques that are not required in every system in development but are relatively new to the field of radar. The current volume, Radar Applications, cites specific examples of the use of basic principles and advanced techniques. It is interesting to note that many of the signal-processing techniques in use today were first discussed in the early (World War II era) series prepared at the MIT Radiation Laboratory.1 The techniques were known, but available signal-processing technology

1 This refers to a twenty-one-book series of topics related to radar technology titled MIT Radiation Laboratory Series, McGraw Hill Book Company, New York, NY, 1948.

Preface did not support implementation until modern digital signal-processing equipment became available.

Acknowledgements As editors for this volume, we are very grateful to the publisher, Dudley Kay, for his enduring support and encouragement. Special thanks also go to Brent Beckley for all of his efforts on the sales and marketing front. We are also grateful to Dudley and Brent for gathering and managing the unusually numerous volunteer reviewers whose participation as a ‘‘community effort’’ over the course of the three-volume series has been remarkable and inspiring. Most important, though, we remain thankful to our families for their patience, love, and support as we prepared materials, revised, reviewed, coordinated, and repeated. This book, like the others, represents time away from the ones we love. We thank them for their understanding, kindness, and support.

To Our Readers We hope the reader will enjoy this book! Radar is and will continue to be an immensely exciting and diverse field of engineering. Please report errors and refinements. We know from the publication of the first two volumes that even the most diligently reviewed and edited book is bound to contain errors in the first printing. It can be frustrating to see such errors persist, even in many subsequent printings. We continue to appreciate SciTech Publishing’s commitment to correct errors and enhance the book with each printing. So, it remains a ‘‘community effort’’ to catch and correct errors and improve the book. You may send your suspected errors and suggestions to: [emailprotected] This email will reach us and SciTech concurrently so we can confer and confirm the modifications gathered for scheduled reprints. You are always welcome to contact us individually as well. Bill Melvin Atlanta, GA Jim Scheer New Bern, NC

xiii

Reviewer Acknowledgements SciTech Publishing – IET gratefully acknowledges the manuscript reviewing efforts from the following members of the international radar and electronic warfare community. Refinements to the book’s content and expression for the benefit of all readers represent the blessing of ‘a community effort’. Ron Aloysius, Northrop Grumman Corporation, USA Edward R. Beadle, Harris Corporation, USA Lee Blanton, General Atomics Corporation, USA Neal Brune, Esterline Defense Technologies, USA Kernan Chaisson, Captain, USAF (retired) I-Ting Chiang, Qualcomm, USA Jean Yves Chouinard, Universite Laval, Canada Patrick Dever, Northrop Grumman Corporation, USA John Erickson, Wright-Patterson Air Force Base, USA Phillip Fitch, University of South Australia, Australia Riccardo Fulcoli, Selex ES, Italy Gaspare Galati, Universita` di Roma Tor Vergata, Italy Frank Gekat, Selex Systems Integration, Germany Martie Goulding, MDA Corporation, Canada Hugh Griffiths, University College London, UK Stephen Harman, QinetiQ, UK Stephen Hogue, Harris GCSD, USA Michael Inggs, University of Cape Town, South Africa Stephane Kemkemian, Thales Airborne Systems, France Peter Knott, Fraunhofer Institute for High Frequency Physics and Radar Techniques, Germany Thodoris G. “Ted” Kostis, University of the Aegean, Greece Anthony Leotta, ADL Associates, USA David Long, Brigham Young University, USA John Milan, Consultant, USA Lee Moyer, Technology Service Corporation, USA Karl-Erik Olsen, Norwegian Defence Research Establishment, Norway A. M. (Tony) Ponsford, Raytheon Canada Ltd., Canada Pinaki S. Ray, University of Adelaide, Australia Earl Sager, Consultant, USA John SantaPietro, The MITRE Corporation, USA Margaret M. “Peggy” Swassing, 412th Test Engineering Group (Edwards AFB), USA Firooz Sadjadi, Lockheed Martin Corporation, USA John Sahr, University of Washington, USA Alexander Singer, Thales Group, Canada Koen van Caekenberghe, HiSilicon, Belgium

xv

Editors and Contributors Volume Editors Dr. William Melvin Volume Editor-in-Chief and Multiple Chapter Author William Melvin is Director of the Sensors and Electromagnetic Applications Laboratory at the Georgia Tech Research Institute and an Adjunct Professor in Georgia Tech’s Electrical and Computer Engineering Department. His research interests include systems engineering, advanced signal processing and exploitation, and high-fidelity modeling and simulation. He has authored more than 180 publications in his areas of expertise and holds three patents on adaptive radar technology. Among his distinctions, Dr. Melvin is a Fellow of the IEEE, with the follow citation: ‘‘For contributions to adaptive signal processing methods in radar systems.’’ He received the PhD, MS, and BS (with High Honors) degrees in Electrical Engineering from Lehigh University. Mr. James A. Scheer Associate Volume Editor and Chapter 1 Author Jim Scheer has forty years of hands-on experience in the design, development, testing, evaluation, and analysis of radar systems. He currently consults and works part-time for GTRI and teaches radarrelated short courses. He began his career with the General Electric Company (now Lockheed Martin Corporation), working on the F-111 attack radar system. In 1975, he moved to GTRI, where he worked on radar system applied research until his retirement in 2004. Mr. Scheer is an IEEE Life Fellow and holds a BSEE degree from Clarkson University and the MSEE degree from Syracuse University. His primary interests are in the area of coherent radar performance prediction and evaluation.

Chapter Contributors Mr. Chris Baker Chris Baker is the Ohio State Research Scholar in Integrated Sensor Systems at The Ohio State University. Until June 2011, he was the Dean and Director of the College of Engineering and Computer Science at the Australian National University (ANU). Prior to this, he held the ThalesRoyal Academy of Engineering Chair of intelligent radar systems based at University College London. He has been actively engaged in radar xvii

xviii

Editors and Contributors systems research since 1984 and is the author of more than 200 publications. His research interests include coherent radar techniques, radar signal processing, radar signal interpretation, electronically scanned radar systems, radar imaging, and natural and cognitive echo-locating systems. He is the recipient of the IEE Mountbatten premium (twice) and the IEE Institute premium, and he is a Fellow of the IET. He is a Visiting Professor at the University of Cape Town, Cranfield University, University College London, Adelaide University, Wright State University, and Nanyang Technical University. Mr. Bill Ballard William Ballard is a Senior Research Associate at the Georgia Tech Research Institute Sensors and Electromagnetic Applications Laboratory. He is also the course director of the popular Georgia Tech Professional Education Airborne Fire Control Systems short course. He is a retired U.S. Navy Commander with more than 3,500 hours and 912 traps in the A-6 Intruder. He has served on the faculty at the NATO School (SHAPE) in Oberammergau, Germany where he taught NATO Maritime Operations, Conventional Weapons Employment, and Naval NBC Defense. Both his bachelor’s and master’s degrees in Mechanical Engineering and Management Science are from Georgia Tech. Mr. Melvin L. Belcher Mel Belcher is a Principal Research Engineer at Georgia Tech Research Institute (GTRI). He has worked in the development and analysis of phased array radars systems for more than three decades. He has focused on air- and missile-defense applications and has also contributed to airborne radar and space surveillance radar efforts. His professional interests include systems engineering, active electronically scanned arrays, and signal and data processing. He currently serves as the Technical Director of the Sensors Knowledge Center within the Missile Defense Agency. He founded and led the Air and Missile Defense Division at GTRI. He served as Chief Engineer for Radar Futures at Northrop Grumman Mission Systems from 2005 through 2010. He received the MSEE from Georgia Institute of Technology and the BEE from Auburn University. Mr. Lee Blanton Lee Blanton is a radar engineer with General Atomics Aeronautical Systems, Inc., where he supports development of radars for unmanned aerial vehicles (UAVs). His thirty-five-year career in industry has focused primarily on radars for airborne, missile-borne, and spaceborne applications with additional work in the areas of satellite communication and electronic warfare systems. His spaceborne radar experience includes design studies for the proposed Venus Orbiting Imaging Radar and its successor, the Magellan Venus Radar Mapper, as well as concept studies for spaceborne radars for the Air Defense Initiative (ADI), the Strategic Defense Initiative (SDI), and imaging radar applications.

Editors and Contributors Dr. Mark E Davis Mark Davis has forty-five years of experience in government and industry in developing technology and systems for radar and electronic systems. After retirement in 2008, he established MEDavis Consulting as a sole proprietorship to serve the radar science and technology community. He served at DARPA as Deputy Director Information Exploitation Office from 2006 to 2008. Prior to this assignment, he was the Technical Director for Air Force Research Laboratory Space Based Radar Technology (1998–2006) and Program Manager in DARPA Information Systems Office for Counter CC&D technologies (1995– 1998). His interests are in radar and microwave system design, phased array antennas, and adaptive signal processing. Dr. Davis is a Life Fellow of the IEEE and Military Sensing Symposia and is Chair of the IEEE Radar Systems Panel. He has a PhD in Physics from The Ohio State University and bachelor’s and master’s degrees in Electrical Engineering from Syracuse University. Dr. Antonio De Maio Dr. Antonio De Maio was born in Sorrento, Italy, on June 20, 1974. He received the DrEng degree (with honors) and the PhD degree in information engineering, both from the University of Naples Federico II, Naples, Italy, in 1998 and 2002, respectively. From October to December 2004, he was a Visiting Researcher with the U.S. Air Force Research Laboratory, Rome, New York. From November to December 2007, he was a Visiting Researcher with the Chinese University of Hong Kong. Currently, he is an Associate Professor with the University of Naples Federico II. His research interest lies in the field of statistical signal processing, with emphasis on radar detection, optimization theory applied to radar signal processing, and multiple-access communications. Dr. De Maio is an IEEE Fellow and the recipient of the 2010 IEEE Fred Nathanson Memorial Award as the young (less than forty years of age) AESS Radar Engineer 2010 whose performance is particularly noteworthy as evidenced by contributions to the radar art over a period of several years, with the following citation for ‘‘robust CFAR detection, knowledge-based radar signal processing, and waveform design and diversity.’’ Dr. Alfonso Farina Alfonso Farina received the doctorate degree in electronic engineering from the University of Rome (I), Italy, in 1973. In 1974, he joined Selenia, now SELEX Electronic Systems, where he has been a Manager since May 1988. He was Scientific Director in the Chief Technical Office. He was the Director of the Analysis of Integrated Systems Unit, Director of Engineering in the Large Business Systems Division, and Chief Technology Officer of the Company (SELEX Sistemi Integrati). Today he is Senior Advisor to CTO of SELEX ES. From 1979 to 1985, he has also been a Professor of radar techniques with the University of Naples. He is the author of more than 500 peer-reviewed technical publications as well

xix

xx

Editors and Contributors as the author of books and monographs. He has been nominated Fellow of IEEE, international fellow of the Royal Academy of Engineering, United Kingdom, and Fellow of EURASIP. He has been appointed member in the Editorial Boards of IET Radar, Sonar and Navigation (RSN) and of Signal, Image, and Video Processing Journal (SIVP). He has been the General Chairman of the IEEE Radar Conference, 2008. He is a Fellow of the Institution of Engineering and Technology (IET), United Kingdom. He is also the recipient of the 2010 IEEE Dennis J. Picard Gold Medal for Radar Technologies and Applications with the following citation: ‘‘For continuous, innovative, theoretical and practical contributions to radar systems and adaptive signal processing techniques.’’ Mr. E. F. Greneker Mr. E. F. Greneker was employed by Georgia Tech Research Institute (GTRI) for thirty-three years before his retirement as a Principal Research Scientist. He was responsible for the establishment of the GTRI Severe Storms Research Center (SSRC) and served as the founding Director of the SSRC. During his career with GTRI he directed more than sixty major sponsored research projects for many U.S. government agencies and the military services. Many of these sponsored projects related to the use of radar for national security purposes. Other projects included using radar to track insects, and police radar. He has authored more than 85 papers, journal articles, and the chapter on police radar in this book. He is a Senior member of the IEEE. He holds five U.S. patents, with others and two as sole inventor. While employed by GTRI, he consulted to government agencies through his consulting firm, Greneker and Associates, Inc. After retiring from GTRI, he started his own business, RADAR Flashlight, LLC (RFLLC). RFLLC performed research for the Defense Advanced Research Projects Agency and the U.S. Army on topics relating to detecting humans through a wall using radar. RFLLC also performed research for the U.S. Air Force on radar detection of moving targets from an unmanned aerial vehicle. Mr. Greneker’s current interests include remote sensing, both optical and radar, passive radar, radar applications for highway safety, and radar used for security purposes. Mr. Hugh Griffiths Hugh Griffiths was educated at Hardye’s School, Dorchester, and Keble College, Oxford University, where he received the MA degree in Physics in 1978. He also received the PhD (1986) and DSc (Eng) (2000) degrees from the University of London. He holds the THALES/Royal Academy of Engineering Chair of RF Sensors at University College London. From 2006 to 2008, he served as Principal of the Defence College of Management and Technology, at the Defence Academy, Shrivenham. From 1982 to 2006, he was with University College London as Head of the Department of Electronic and Electrical Engineering from 2001 to 2006. His research interests include radar sensor systems and signal processing (particularly synthetic aperture radar and bistatic

Editors and Contributors and multistatic radar and sonar) as well as antennas and antenna measurement techniques. He has published more than 400 papers and technical articles on these subjects. Professor Griffiths was awarded the IET A F Harvey Prize in 2012 for his work on bistatic radar. He has also received the IERE Lord Brabazon Premium Award in 1984, the IEE Mountbatten and Maxwell Premium Awards in 1996, and the IEEE Nathanson Award in 1996. He served as President of the IEEE AES Society for 2012–13. He has been a member of the IEEE AESS Radar Systems Panel, which he chaired from 2006 to 2009, and is Editor-in-Chief of the journal IET Radar, Sonar, and Navigation. He was Chairman of the IEE International Radar Conference RADAR 2002 in Edinburgh, United Kingdom. He also serves on the Defence Scientific Advisory Council for the U.K. Ministry of Defence. He is a Fellow of the IET, Fellow of the IEEE, and in 1997 was elected to Fellowship of the Royal Academy of Engineering. Mr. Stephane Kemkemian Stephane Kemkemian graduated in Aerospace Engineering from ISAE, Toulouse, France. He began his career at Thales working on RDY and RBE2 radar prototypes (radars for the Mirage-2000 and Rafale fighters, respectively). He is now senior expert with the Technical Directorate of Thales Airborne Systems. He holds around thirty patents and is the author of about twenty papers. He is senior member of the French SEE and founding member of the IEEE AESS French chapter. Dr. Richard C. Liu Richard C. Liu received his BS, MS, and PhD degrees in radio engineering from Xi’an Jiaotong University, Xi’an, China, in 1982, 1984, and 1988, respectively. Since 1988, he has been with the Department of Electrical and Computer Engineering, University of Houston, Houston, TX, where he is currently a Professor and the Director of the Well Logging Laboratory and the Subsurface Sensing Laboratory. His research areas include resistivity well logging, tool simulation, tool hardware design, electromagnetic telemetry systems, ground-penetrating radar, sensor technology, wireless telecommunication systems, shortrange radio, and RF and microwave circuit design. Dr. Liu has published more than 160 technical papers in these areas. Dr. Liu is a senior member of IEEE, member of the Society of Professional Well Logging Analysts, the Environmental and Engineering Geophysics Society, and the Society of Core Analysts. Mr. Aram Partizian Aram Partizian is a Senior Research Scientist at GTRI, where he contributes to the design, development, and field-testing of advanced radar electronic warfare technologies. He has more than thirty years of experience in radar and the electronic warfare field, including software and system engineering roles at Raytheon Company prior to joining Georgia Tech. He earned a BA in Physics from Oberlin College in 1977.

xxi

xxii

Editors and Contributors Mr. Samuel Piper Samuel O. Piper is a GTRI Fellow and Principal Research Engineer, and since 2002 has served as Chief of the Radar Systems Division in the Sensors and Electromagnetic Applications Laboratory in GTRI. He has more than forty years of experience in radar systems engineering and analysis. He serves as coordinator for the Georgia Tech Principles of Continuous Wave (CW) Radar short course. He earned a master’s degree in EE from Georgia Tech. Mr. John Porcello John Porcello is a Senior Research Engineer for the Georgia Tech Research Institute (GTRI). John designs, develops, and implements digital signal processing (DSP) algorithms in FPGAs for a wide range of applications, including radar and communications. John has more than twenty years of experience as an engineer and is a Senior Member of the IEEE and a private pilot. Mr. Jay Saffold Jay Saffold is the Chief Scientist for RNI and has more than twenty years of engineering experience in both military and industry research in RF tags, virtual reality, digital databases, soldier-tracking systems, millimeter wavelength (MMW) radar, multimode (MMW and optical) sensor fusion, fire-control radar, electronic warfare, survivability, signal processing, and strategic defense architecture. He lectures annually for GTRI on remote sensing and signal processing. He has authored or coauthored more than 104 technical papers and reports. He holds a BSEE degree from Auburn University. Dr. Luca Timmoneri Luca Timmoneri is a Vice President of SELEX ES, where he is currently Chief Technical Officer of the Land and Naval Division. His working interests span from the area of synthetic aperture radar, to radar STAP, to detection and estimation with application to tridimensional phased array radar, to parallel-processing architectures. Dr. Timmoneri is the author of several peer-reviewed papers (also invited). He is the coauthor of three tutorials presented at International IEEE radar conferences. He received the 2002, 2004, and 2006 AMS (now SELEX ES) CEO Award for Innovation Technology; the 2003 AMS (now SELEX Sistemi Integrati) MD Award for Innovation Technology; the 2004 Finmeccanica Innovation Award; and the 2013 Oscar Masi Award for industrial innovation of the Italian Association for Industrial Research. Mr. John Trostel John Trostel is a senior research scientist and Director of the Severe Storms Research Center at the Georgia Tech Research Institute (GTRI). His fields of specialization include the meteorology of severe storms, development of data-acquisition and analysis systems, effects of meteorological phenomena on MMW propagation and backscatter, and

Editors and Contributors general physics and meteorological expertise. He was part of a GTRI team tasked to support the FAA in the development of a Multifunction Phased Array Radar (MPAR) system and was involved in investigations of MMW backscatter characteristics of snow-covered ground, atmospheric acoustics, and underwater sonar development. He is an active member of the American Meteorological Society, National Weather Association, American Geophysical Union, and American Physical Society.

xxiii

CHAPTER

Radar Applications

1

William L. Melvin, Ph.D., and James A. Scheer, Georgia Institute of Technology, Atlanta, GA

Chapter Outline 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9

Introduction . . . . . . . . . . . . . . . . . Historical Perspective. . . . . . . . . . Radar Measurements . . . . . . . . . . Radar Frequencies . . . . . . . . . . . . Radar Functions. . . . . . . . . . . . . . U.S. Military Radar Nomenclature Topics in Radar Applications . . . . Comments . . . . . . . . . . . . . . . . . . References. . . . . . . . . . . . . . . . . .

1.1

........ ........ ........ ........ ........ ........ ........ ........ ........

........ ........ ........ ........ ........ ........ ........ ........ ........

....... ....... ....... ....... ....... ....... ....... ....... .......

........ ........ ........ ........ ........ ........ ........ ........ ........

...... 1 ...... 2 ...... 5 ...... 6 ...... 8 ...... 9 . . . . . . 10 . . . . . . 14 . . . . . . 15

INTRODUCTION

Radio detection and ranging (radar) involves the transmission of an electromagnetic wave to a potential object of interest, scattering of the wave by the object, receipt of the scattered energy at the receive site, and signal processing applied to the received signal to generate the desired information product. Originally developed to detect enemy aircraft during World War II, radar has through the years shown diverse application, not just for military consumers, but also for commercial customers. Radar systems are still used to detect enemy aircraft, but they also keep commercial air routes safe, detect speeding vehicles on highways, image polar ice caps, assess deforestation in rain forests from satellite platforms, and image objects under foliage or behind walls. A number of other radar applications abound. This book is the third in a series. Principles of Modern Radar: Basic Principles, appearing in 2010, discusses the fundamentals of radar operation, key radar subsystems, and basic radar signal processing [1]. Principles of Modern Radar: Advanced Techniques, released in 2012, primarily focuses on advanced signal processing, waveform design and analysis, and antenna techniques driving tremendous performance gains in radar system capability [2]. This third text, Principles of Modern Radar: Radar Applications, combines the developments of Basic Principles and Advanced Techniques to illustrate a myriad of radar applications.

1

2

CHAPTER

1

Radar Applications

Principles of Modern Radar: Radar Applications is comprised of three sections: ●

Tactical Radar, covering continuous wave radar, with application to missile seekers and other low-cost radar needs; millimeter wave radar, used in areas such as battlefield fire-control systems and automotive radar; fire-control radar principles; airborne pulse Doppler radar, the heart of airborne interceptor radar; multifunction radar used to search, track, and engage airborne targets and employing sophisticated and costly phased array antennas, processing software, and resource management; and ballistic missile defense radar. Intelligence, Surveillance, and Reconnaissance, including early warning detection of aircraft and missiles preceding handoff to a tracking radar; surface moving target indication, used to detect and monitor targets on Earth’s surface; and spaceborne surveillance used to remotely monitor Earth resources, cultural sites, and military facilities. Specialized Applications, including passive radar, which uses noncooperative sources of illumination and receivers displaced a considerable distance from the various transmit sites; air traffic control radar; weather radar; foliage-penetrating radar; ground- and materials-penetrating radar; and police radar.

Individual chapters discuss the aforementioned topics within these three sections in further detail, identifying key considerations and the practical application of radar technology, principles, and techniques to accomplish the specific radar objective: detecting, locating, and tracking targets moving on Earth’s surface; imaging a stationary target under foliage; detecting approaching or receding targets from an airborne pulse Doppler radar; detecting and tracking ballistic missiles from large, ground-based phased array radar; protecting ground troops from mortar attack using mobile, counterbattery surface radar; and so on.

1.2

HISTORICAL PERSPECTIVE

The earliest radar developments appear to have taken place independently in a number of countries. World War II accelerated the development of radar to address the direst of situations. That military application has served as a primary motivation for radar technology development complicates an exposition on its history due to sundry requirements for secrecy. Consequently, spirited debate amongst radar developers over who deserves acclaim for certain innovations is not uncommon. Reference [3] provides a remarkable overview of the earliest beginnings of radar. The possibility of a system to detect objects based on reflected electromagnetic waves dates to the 19th century and the work of Heinrich Hertz, with James Clerk Maxwell’s work on electromagnetism suggesting this possibility. Other great minds invariably associated with the earliest beginnings of radar include Christian Hulsmeyer, Nikola Tesla, Guglielmo Marconi, Sir Robert Wattson-Watt, and Hoyt Taylor. As [3] discusses, highly protected programs to develop radar took place leading up to and during World War II in a number of countries, including England, France, Germany, Japan, Canada, and the United States. Robust radar programs were further known to exist in the Soviet Union, Italy, and the Netherlands.

1.2

Historical Perspective

The detection of air raids was of paramount importance during World War II. Generally, surface-based radars, such as the British Chain Home radar system [4], were developed for this purpose. These original surveillance radars provided an early warning function so citizens could take shelter and service personnel could launch interceptor aircraft. The interceptors similarly required radar to acquire and engage enemy bombers and provide self-protection from enemy escort aircraft. Early warning and fire-control radar were also necessary for naval shipboard protection. World War II applications solidified the need for microwave transmitters and receivers and pulsed waveforms. As pointed out in [4], this period of extensive innovation involved the efforts of multiple researchers and engineers, resulting in radar having no single lineage, but a collection of forefathers. The earliest radar experiments involved continuous waveforms and bistatic configurations to achieve sufficient isolation between transmitter and receiver [4, 5]. The technology available at the time could only support detection; range information was not available to the operator. Moreover, many of these initial investigations involved longer wavelengths – in the vicinity of 60 cm or greater. A requirement for range information and improved spatial accuracy led to microwave developments and pulsed radar modes. For years beyond World War II, noncoherent pulsed radar systems were used for a number of important applications. A coherent radar employs a stable, coherent oscillator to transmit and receive signals. In this manner, the radar keeps track of the phase of the receive signal over time. A timevarying phase leads to a frequency shift in the receive signal. If the range between the radar and the object of interest is changing, the time it takes the signal to propagate to the object and return to the radar is tðtÞ ¼ 2rðtÞ=c, where rðtÞ is the time-varying range and c is the velocity of propagation (nominally, the speed of light). The corresponding phase is fðtÞ ¼ wtðtÞ, where w is frequency in radians. Frequency is the time-derivative of phase, _ fðtÞ ¼ w@tðtÞ=@t. Suppose rðtjt ¼ nTÞ ¼ r0 þ nDr, with T the sample time, n the sample index, r0 the initial range, and Dr the constant change in range between sample times resulting from a constant velocity target. The corresponding derivative of the phase _ function is fðtjt ¼ nTÞ ¼ ð4p=lÞðDr=DtÞ, with Dt the change in time; we recognize _ ¼ nT Þ ¼ wd ¼ ð4 p=lÞ vr Dr=Dt as the radial velocity (or range–rate), vr , and fðtjt (or, fd ¼ 2vr =l in Hz) as the well-known Doppler shift [4, 6]. The ability to take advantage of the target Doppler shift was revolutionary, providing the radar with additional information on target motion and enabling a mechanism to better separate target returns from those of background clutter due to reflections from Earth’s surface or even from weather phenomenon. Thus, the extensive development of coherent radar systems followed the major accomplishments of the World War II era and occupied the minds of radiofrequency scientists and engineers for subsequent decades. The pulsed Doppler mode is the cornerstone of modern radar technology, integral to surface and aerospace military radar systems. Pulsed Doppler radar has important civilian and commercial applications, permeating everyday life in the form of television weather newscasts with detailed radar weather maps and air traffic control radar making the skies safe for travelers of all types. Coherent continuous wave radars are also important, providing target Doppler information for applications ranging from missile engagement to police traffic surveillance. Coherency also makes all-weather terrain and stationary target mapping possible via a technology known as synthetic aperture radar (SAR) [2, 6–8]. SAR was invented

3

4

CHAPTER

1

Radar Applications

in the 1950s, with Carl Wiley of Goodyear Aircraft Company viewed as its originator, and multiple parties greatly contributing to its development. The primary objective of SAR is to create a high-resolution map of the scene reflectivity; the resulting product has image-like quality and is generally interpreted by a trained analyst. In its most basic form, SAR uses knowledge of the collection geometry, generating a matched filter tailored to the phase history of a particular, resolvable, stationary scatterer of interest. As previously discussed, the phase history is fðtÞ ¼ wtðtÞ, where in this case the change in range over time typically leads to a nonlinear characteristic for tðtÞ, and consequently a response comprised of time-varying frequency. The SAR is built and deployed in such a way that ideally the various scatterers possess unique phase histories, though practically there are basic limitations affecting the quality of the reflectivity estimate. Image-formation processing is the series of steps applied to the phase history data to generate the SAR map. SAR has played important military roles in areas such as nuclear arms treaty monitoring and battlefield surveillance, preparation, and damage assessment. Meeting these stringent and critical applications required extraordinary effort to achieve pristine coherency over relatively long periods of time – data are collected over periods of hundreds of milliseconds to tens of minutes or more, a duration required to traverse sufficient viewing angle to achieve a desired cross-range resolution – and conceive computationally feasible approaches to approximate the matched filter condition. Indeed, system coherency and signal-processing algorithm development have served as hallmarks of SAR technology development. Early SAR image formation used optical signal-processing methods, with digital signal-processing techniques replacing the former after a relatively extended period of time needed for available technology to sufficiently advance. With some delay, civil applications of SAR emerged, including Earth resources monitoring, polar ice cap monitoring, and extraterrestrial planetary exploration. Over the past twenty years or so, the radar community has significantly focused on radar subsystem hardware improvement, signal-processing algorithm development and implementation, and diverse applications. The development of phased array radar has been a major undertaking and a critical step in radar deployment for air and missile defense and multimode airborne radar systems [9]. Advances in computing technology have made digital beamforming (DBF) and space-time adaptive processing (STAP) possible [2, 9–11]. DBF and STAP are key elements in radar electronic protection, superior clutter mitigation techniques, and advanced concepts such as passive radar where DBF makes ‘‘pulse chasing’’ feasible [5]. Radar’s diverse applications made possible through technology maturation include through-the-wall radar for law enforcement support; the detection, location, and characterization of dismount targets (persons of interest traversing Earth’s surface) from airborne radar [12]; remote sensing of ocean currents; border surveillance; gait analysis for threat monitoring (e.g., detection of a perimeter breach by unauthorized personnel) and medical diagnosis (e.g., assessment of indicators of traumatic brain injury); automotive radar for intelligent highways; and the development of low-cost passive surveillance radar hosting off of commercial communications broadcasts [13]. Radar has proven its importance to society. As such, radar development and implementation has generally received favorable treatment under situations of competing interest. An emerging conflict over spectrum allocation among users of the electromagnetic spectrum will intensify, leading radar developers to innovate and conceive new technology and capabilities [14]. In addition to spectrum, energy is

1.3

Radar Measurements

placing pressure on radar: The proliferation of wind farms as an alternative energy source creates a whole new class of interference requiring mitigation to ensure effective radar performance. Radar will also be asked to solve new and challenging problems, such as identification of humans in emergency management situations resulting from such natural phenomena as earthquakes; the detection of small vessels traversing the littoral zone expanse; and the beneficial exploitation of multipath in urban settings to enable non–line-of-sight radar detection and tracking of objects [15]. This book summarizes and puts into perspective a select number of important and modern radar applications, as well as the requisite constituent technology. As such, it builds on the exposition set forth in the first two volumes of the Principles of Modern Radar series [1, 2].

1.3

RADAR MEASUREMENTS

Radar operation requires an active source of illumination. Monostatic and cooperative bistatic radar use a coordinated transmitter. Noncooperative bistatic radar exploits the transmissions from other electronic systems, including radio towers, communication transmit antennas, and other radars. Cooperative systems attempt to tailor the transmit waveform to the extent possible to maximize important radar quality measures, such as signal-to-noise ratio (SNR), signal-to-interference-plus-noise ratio (SINR), range resolution, target class separation, and resilience to radiofrequency interference (RFI). The radar generates its product based on target-induced modulation of the reflected waveform. Radar design allows access to the following primary measurements: ●

Fast-time – collected at the analog-to-digital converter rate, these voltages correspond to sampling in the range dimension. Slow-time – collected at the pulse repetition interval (PRI), the corresponding voltages are the pulse-to-pulse measurements for a given range cell. The Fourier transform of slow-time is Doppler. Spatial – samples generated at the output of a multichannel or multibeam receive antenna, where each channel or beam has its own receive chain. Angle information follows from the Fourier transform of the spatial channel measurements; the inverse transform of the multibeam output restores spatial sample information. The measured angle corresponds with azimuth, elevation, or cone, where cone is an ambiguous measurement related to a specific direction cosine in the antenna coordinate system. Polarimetric – consists of two basic forms, dual-polarization and quad-polarization. In dual-polarization, the transmit polarization is fixed and the receive antenna collects orthogonal polarizations (e.g., the transmitter sends out a vertically polarized wave, and the receiver collects both vertical and horizontal polarizations). Quadpolarized operation requires the transmitter to interleave transmissions of orthogonal polarizations, and the receiver simultaneously collects two orthogonal polarizations as in the dual-polarized case. Multipass – the radar can collect data at a common operating frequency, polarization, and bandwidth over distinct orbits and then process the data to look for scene changes. When the processing is coherent from pass to pass, the mode is called

5

6

CHAPTER

1

Radar Applications

coherent change detection (CCD); naturally, noncoherent change detection operates on magnitude-only data from pass to pass. Change detection makes it possible to detect subtle changes in the scene, such as the presence of tire tracks on a dirt road or areas of trampled grass. It is the purpose of the radar signal processor to operate on the radar measurements and generate the radar data product. A data signal processor, such as a tracker, operates on this output to assist the operator or analyst in interpreting events. Regarding radar measurements, it is worth pointing out the difference between monostatic, bistatic, and multistatic systems [2]. The radar transmitter and receiver are collocated in monostatic radar. The bistatic configuration employs transmit and receive sites separated by an appreciable distance [5]; the distance is not precisely defined, but it is instructive to consider the bistatic configuration one in which target and clutterscattering phenomenology are distinct from the monostatic case, and hence include different information content. A cooperative bistatic system controls its illumination source, whereas a noncooperative bistatic system employs illuminators of opportunity. Multistatic radar merges data from multiple bistatic nodes and can yield substantially enhanced geolocation performance resulting from the combination of the diverse target measurements [19]. Invariably, radar applications involve collecting and exploiting distinct measurements to achieve a given mission objective. Different measurement domains enable the radar to better differentiate a desired target from interference and other potential targets. At times, practical considerations – cost, deployment issues, etc. – affect the measurement domains collected by the radar.

1.4

RADAR FREQUENCIES

Radar operating frequency is chosen based on a number of considerations. Important trade factors include but are not limited to the following. ●

Spatial resolution: For a fixed aperture size, beamwidth is proportional to l=La;m , where l is wavelength and La;m is the aperture length in the mth dimension. Propagation: Lower frequencies propagate farther and are used in very long range surveillance systems. As frequency increases, so does atmospheric attenuation due to water vapor, rainfall, and other weather effects as well as from dust and suspended particulates [16]. Materials penetration: Radar systems that must find targets under foliage, behind walls, under canopies, or below soil favor lower frequency operation. Foliagepenetrating (FOPEN) radar systems typically operate at frequencies from several tens of megahertz up to 1 GHz; ultrahigh frequency (300 MHz to 1 GHz) is a popular choice, trading off attenuation for resolution. Through-the-wall radar favors L-band (1–2 GHz) as a good trade between attenuation through the wall, resolution, and aperture size. Electromagnetic interference/electromagnetic compatibility (EMI/EMC): The characteristics of spectrum use in the vicinity of the radar siting or operating environment influence frequency selection. For example, placing a radar in the vicinity of a

1.4

Radar Frequencies

high-power communications transmitter influences frequency selection and the general system design. ●

● ●

Electronics: The availability and cost of electronic components at a given frequency influence the design. There are many radar systems built at X-band, for instance, leading to lower-cost electronics than at Ku-band, making it more challenging to justify Ku-band designs without other compelling factors. Target properties: Target phenomenology varies with frequency selection [16, 17]. Fractional bandwidth limitations: High resolution requires wider waveform bandwidth and design consideration to accommodate dispersion and hardware mismatch effects. Generally, instantaneous bandwidths drive the system design up to higher operational frequency as a means of simplification. Radiofrequency interference: Radar frequency may be selected to avoid operating in a band covered by jamming systems [2, 18].

Table 1-1 summarizes the radar frequency operating bands. Specific frequency allocations for radar are designated by governing bodies: the International Telecommunications Union (ITU) in particular, with coordination among other national agencies. Example applications for the various frequencies are given in Table 1-1. The nomenclature relates to the function. For example: ● ●

the ‘‘L’’ in ‘‘L-band’’ refers to long range application; Ku is ‘‘K under’’ and Ka is ‘‘K above,’’ respectively, due to their frequency ranges relative to K-band;

TABLE 1-1

¢

Radar Frequency Bands

Frequency

Range

High frequency (HF)

3–30 MHz

Very high frequency (VHF) Ultrahigh frequency (UHF) L-band S-band C-band X-band Ku-band K-band Ka-band Millimeter wave (mmw)

Example Application(s)

Ground-penetrating radar, over-the-horizon radar (OTHR), very long range surveillance radar 30–300 MHz Foliage-penetrating radar, very long range surveillance radar 300–1,000 MHz Foliage-penetrating radar, airborne surveillance radar, long range ballistic missile defense radar 1,000–2,000 MHz Weapons location radar, air traffic control radar, long range surveillance radar 2,000–4,000 MHz Naval surface radar, weapons location radar, weather radar 4,000–8,000 MHz Weather radar 8,000–12,000 MHz Fire-control radar, air interceptor radar, groundmapping radar, ballistic missile–tracking radar 12,000–18,000 MHz Air-to-ground SAR and surface-moving target indication 18,000–27,000 MHz Limited due to absorption 27,000–40,000 MHz Missile seekers, close-range fire-control radar 40,000–300,000 MHz Fire-control radar, automotive radar, law enforcement imaging systems, airport scanners, instrumentation radar

7

8

CHAPTER

1

Radar Applications

the ‘‘X’’ in X-band stands for ‘‘X marks the spot,’’ due to the common use of this frequency for fire-control systems (some suggest that the X is the roman numeral representing 10, the approximate center frequency in GHz for the X-band); and ‘‘C-band’’ is a ‘‘compromise’’ between a selection of X-band and S-band.

The radar center wavelength is given as lo ¼ c=fo , where fo is the center frequency. Wavelengths on the order of a millimeter technically start just slightly above 30 GHz.

1.5

RADAR FUNCTIONS

All radar systems operate on the same physical principle: an active source illuminates a target, a receiver then collects scattered target energy, and a processor generates the radar product (e.g., dots on a screen representing target detections or a synthetic aperture radar image). From this basic concept of radar operation arise different radar functions. Radar mode design implements variants of these core functions: search, track, and recognition. In general, the purpose of the core radar functions falls into one of two primary categories, as given in [2]: ●

moving target indication (MTI), with subsequent steps to estimate target motion and type, perhaps followed by a tracker to refine target position and velocity estimates and predict where the target will next appear; or radar imaging, the process of collecting data, estimating radiofrequency reflectivity over the local coordinates of interest, and then mapping the estimates to a georeferenced framework.

In search, the radar system attempts to acquire targets of interest. Examples include an airborne early warning (AEW) radar scanning the sky for incoming aircraft and an air interceptor (AI) radar scanning for enemy fighter aircraft. In a similar vein, imaging radar typically ‘‘lay down’’ a certain number of beams per specified time interval to collect spotlight SAR data, or scan a certain area on Earth’s surface in stripmap mode with the objective of searching for certain target types; in the former case, the target of interest might be a missile launcher, whilst in the latter scenario the analyst may be trying to identify deforestation or degradation of polar ice caps. Oftentimes, radar systems that implement the search function are called surveillance radar. The surveillance radar may detect the same target multiple times, thereafter tracking the target through the skill of the radar analyst via something tantamount to ‘‘grease pencil markings on a radar display’’ or by feeding radar measurements into an automated tracker; however, the radar continues to search for new targets with a very similar scan pattern and waveform previously employed to generate existing target indications; and, as already suggested, the nuances of correlating these target detections from scan to scan are left to either the analyst or an automated tracker. Radar resources are not diverted upon detecting a given target; rather, if engagement is to occur, the surveillance radar ‘‘hands off’’ the target to a tracking radar. The tracking function involves focusing radar resources more acutely on a particular target or set of targets. The radar dedicates resources to ensure adequate measurements

1.6

U.S. Military Radar Nomenclature

are collected to maintain track quality. Information from the tracker is used to direct the transmit beam to anticipated target locations. For example, an L-band search radar persistently detects an incoming target, thereafter handing off the acquired target to an X-band tracking radar that refines estimates of target state (position, velocity, and possibly acceleration) by frequently collecting target measurements. The product from the tracking radar function is subsequently provided external to the radar system to a command-and-control function. It is possible that a single radar performs both search and track. Moreover, a single radar can, under the appropriate set of constraints, simultaneously implement both functions in what is known as track-while-scan. In track-while-scan, sufficient radar timeline is available so that, between required tracker updates, the radar can allocate its resources to search for new targets or reacquire targets dropped by the tracker. In addition to searching for targets and placing them in track, recognition is another important function. Recognition involves coarsely or finely determining the target type through the following steps: discrimination, classification, and identification. Discrimination bins the target according to level of interest – for example, a potential military target versus generic ground traffic. Classification determines the threat category, such as ground transport, tank, or missile launcher. Identification then narrows the assessment to a particular target class, such as the tank, missile launcher, helicopter, or aircraft model. Different levels of recognition place varying demands on radar resources: discrimination only requires relatively coarse resolution, whereas identification requires greater information and hence higher resolution. These demands force the radar system to modify its operation in a manner distinct from search and track functions. Recognition may take place at the hands of a trained analyst. An overview of automatic target recognition is given in [2].

1.6

U.S. MILITARY RADAR NOMENCLATURE

Radar nomenclature acknowledges many different radar applications. Table 1-2 shows the nomenclature system used to catalog radar systems in the U.S. military. The first letter designates the platform, the second the equipment type, and the third the TABLE 1-2

a

¢

Some Elements of the Joint Electronics Type Designation System (JETDS)a

Platform

Equipment Type

Application

A – Airborne F – Ground fixed M – Ground mobile S – Surface ship T – Ground transportable U – Ground utility B – Underwater G – Ground K – Amphibious P – Man portable

L – Countermeasures P – Radar Y – Processing B – Communications security Q – Sonar W – Armament

G – Fire control or searchlight directing N – Navigation Q – Special or multipurpose Y – Surveillance R – Receive (passive) only S – Detecting, range and bearing, search

The Joint Army–Navy Nomenclature System was previously used to catalog electronic equipment.

9

10

CHAPTER

1

Radar Applications

application; typically, these letters are preceded by the designation ‘‘AN/’’ (for joint service Army–Navy equipment) and followed by the model number. For example: ●

The AN/APY-1 is the radar on the E-3A and E-3B Sentry Airborne Warning and Control System (AWACS). AN/APY-1 reads as ‘‘Army–Navy equipment,’’ airborne platform, radar, surveillance, model number 1. The AN/APY-2 is the radar on the E-3C AWACS and includes a maritime capability. The AN/APG-63 is the radar used on the F-15E fighter aircraft. APG-63 stands for airborne platform, radar, fire-control, model number 63. The AN/TPQ-53 is the Quick Reaction Capability Radar, sometimes called the Enhanced Firefinder radar. TPQ-53 stands for ground transportable, radar, multipurpose, model number 53. The TPQ-53 is a counterbattery radar used to defend ground troops from rocket, artillery, and mortar attack. The TPQ-53 is replacing the TPQ-36 Firefinder radar. The AN/SPY-1 is part of the U.S. Navy’s Aegis Combat System. It is a passive, phased array surveillance radar used to protect the ship from air and missile attack. SPY-1 stands for shipborne platform, radar, surveillance, model number 1.

A vast array of radar systems comprise the U.S. military inventory, covering a tremendously wide range of applications. Moreover, military radar innovation has led to civil and commercial opportunities. This book considers a number of different radar applications, discussing key issues, constraints, and technology resulting in a particular radar capability.

1.7

TOPICS IN RADAR APPLICATIONS

This book is organized into three sections: tactical radar; intelligence, surveillance, and reconnaissance (ISR) radar; and specialized radar applications. Topics assigned to a particular section are done so based on predominant use but may hold broader applicability.

1.7.1 Tactical Radar Tactical radar systems are used to execute an action within a limited timeframe, as opposed to information gathering that indirectly supports future activities. As a military example, tactical radar is used to track and engage an incoming missile. Police radar is used to evaluate speeds of individual vehicles relative to allowable limits and is a civilian safety example. Determination of liquid levels in industrial storage tanks, known as level gauge measurement, is a commercial example. Continuous wave (CW) radar systems imply low-cost, low-complexity radar. These radar systems typically operate at short range, and their applications include missile seekers, altimeters, active protection systems used to direct a kinetic kill response at incoming rockets, police radar, and automotive safety. Chapter 2 discusses CW radar in detail, covering the basic configuration types; CW radar performance issues and analysis; modulated CW waveforms, including the commonly used linear frequency modulated CW (FMCW) waveform; and applications leveraging the benefits of CW radar. Chapter 3 discusses millimeter wave (mmw) radar. As mentioned earlier, the millimeter wave regime technically ranges from 30 GHz to 300 GHz. The shorter

1.7

Topics in Radar Applications

wavelength is appealing for compact radar applications, as would be the case on a missile, in an unmanned aerial vehicle (UAV), or in a personal conveyance. A key benefit of the higher frequency is narrower beamwidth for a fixed aperture size, an important consideration for target engagement and operation in clutter-limited environments. A mmw radar can operate using both CW and pulsed waveforms; current applications tend to favor CW, consistent with the discussion in Chapter 2. Other mmw radar applications include concealed weapon imaging, automotive radar, and autonomous landing systems. In each of these applications, the short wavelength benefits the system application: higher resolution for concealed weapon imaging and autonomous landing; and compact system design with appropriately narrow beamwidth yielding finer angular resolution, as well as improved electromagnetic compatibility, in support of effective automotive radar. As discussed in Chapter 3, interest in mmw radar continues to grow; this interest will lead to improvements in the cost and performance of mmw electronic components. Fire-control systems seek to detect, track, and recognize targets as part of the engagement process. While a number of sensor modalities can be used for fire control, radar proves very appealing, as Chapter 4 discusses, due to its improved range performance and all-weather capability relative to infrared and optical sensors. There is a broad range of fire-control radar systems supporting a number of missions, including air-to-air combat, air-to-ground fixed-site targeting, shipboard protection, and ballistic missile defense. Chapter 4 broadly considers fire-control radar objectives, implementation considerations, and example systems. This information is a good segue into subsequent chapters. Pulse Doppler waveforms are a critical element of current and future radar systems. This waveform is the mainstay of most radar modes; the pulse Doppler waveform is particularly useful since it provides superior transmit-to-receive isolation. While SAR, AEW, and surface moving target indication (SMTI) all use pulse Doppler variants, air-to-air pulse Doppler radar is the focus of Chapter 5. Chapter 5 discusses basic airborne pulse Doppler radar principles and concepts, characterizes target and clutter Doppler properties as seen from an airborne platform, and examines the various pulse repetition frequency (PRF) selections. The design of the antenna subsystem plays a critical role in modern radar capability and sophistication. Multifunction phased arrays offer superlative performance, since a single radar can carry out multiple tasks. As Chapter 6 describes, multifunction phased arrays provide beam agility through fine control of the elements comprising the antenna. Surface air and missile defense radars, such as the AN/TPY-2 Terminal High Altitude Area Defense (THAAD) radar, and airborne radars, such as the AN/APG-81 radar on the F-35 Lightning, provide multifunction capabilities to search, provide track-while-scan on many targets, and support weapons engagement. The multifunction phased array radar rapidly focuses a beam in space, transmits and receives an appropriate waveform for that specific objective, and then rapidly moves the beam electronically to the next dwell position. In addition to leveraging advanced antenna technology, multifunction phased array radar systems require detailed software architectures to manage system resources. Chapter 6 discusses resource management, as well as multifunction phased array design and performance assessment. Ballistic missile defense (BMD) is an important application for multifunction phased array radar. BMD is an extraordinarily challenging problem, dealing with vast detection ranges and targets of lower RCS and higher velocity than typically seen in other applications. The BMD problem is sometimes stated as ‘‘hitting a bullet with a

11

12

CHAPTER

1

Radar Applications

bullet’’ due to its complexity. Chapter 7 describes in detail BMD radar and its corresponding reliance on large, costly, and highly capable phased array radar systems. These phased array radar systems provide exquisite sensitivity and agility to detect, track, and engage ballistic missile targets. Moreover, as Chapter 7 describes, the radar systems comprising the BMD system usually accomplish other important missions as well, including shipboard defense for Aegis BMD, space situational awareness at some of the large ground-based radar sites, and measurement and signature intelligence (MASINT).

1.7.2 ISR Radar ISR radar systems gather information in support of other actions. Examples include the collection of spotlight SAR imagery to determine if activity is taking place in the vicinity of a missile site and detection of troop movement using ground moving target indication (GMTI) radar [20]. Radar systems dedicated to early warning are also considered ISR assets. Early warning served as the original motivation for radar development. The British Chain Home radar is among the earliest early warning radar systems, and it played a pivotal role in the Battle of Britain during World War II. Since these early days, early warning radar systems continue to flourish, and many experts recognize their capabilities as critical to national defense. These early warning radar systems feed into command-andcontrol networks and provide handoff to tracking and engagement radar. Ground-based, shipborne, and airborne variants exist. Chapter 8 focuses on ground-based early warning radar, complementing some of the discussion in Chapter 7 on ballistic missile warning. Discussion in Chapter 8 covers the objectives of early warning radar; antenna, transceiver and electronics, signal processing, tracking, and electronic protection design considerations; and characteristics of fielded early warning radar systems. This chapter provides international exposure to the topic. Chapter 9 covers SMTI radar design and implementation. (GMTI is the most prominent instantiation of SMTI.) SMTI is a radar mode whose fielded history started in the early 1990s with Joint STARS [20]. This chapter discusses the fundamentals of SMTI, including clutter and target modeling, performance measures, system design considerations, and signal-processing requirements. Clutter mitigation is a critical topic in SMTI, and substantial effort is devoted in Chapter 9 to this topic. As described in the chapter, at its very essence, SMTI radar attempts to discriminate the angle-Doppler response of a potential target from the background clutter. The chapter describes an end-to-end detection processing chain and a standard approach to bearing and Doppler estimation. An overview of several critical topics affecting SMTI implementation, such as the impact of heterogeneous clutter on detection performance and requirements for dismount detection, conclude the chapter. Deploying radar on Earth-orbiting satellites is appealing due to the access such platforms provide. In recent years, international interest in developing and deploying satellite-based synthetic aperture radar has exploded. Spaceborne SAR has numerous applications, including remote sensing of natural resources, monitoring of oceans and gulfs, emergency management, and treaty monitoring. Chapter 10 discusses spaceborne SAR. The chapter presents an array of internationally developed SAR systems, exhaustively covers a number of critical design issues and considerations, and describes SAR implementation and performance assessment applied to spaceborne

1.7

Topics in Radar Applications

assets. The chapter summarizes the characteristics of a number of operational spaceborne SAR systems.

1.7.3 Specialized Applications Innovation in radar technology continues. Advances in RF electronics and antenna technology, as well as remarkable improvements in high-performance computing, enable the conception and deployment of numerous new radar capabilities. This section of the book examines the emerging or specialized applications of radar technology. Passive bistatic radar systems exploit ambient signals, such as those from broadcast stations and communications towers, to detect and localize moving targets. Original observations on radar potential were a result of target-induced modulation on noncooperative signals viewed at a receive site; early radar systems were bistatic owing to a requirement to isolate the transmit and receive function, and the history of radar in general and that of the bistatic topology are inseparable. The availability of lower-cost electronic components and computing devices is a key enabler in the design and deployment of passive bistatic radar, and it is a primary reason for an international surge of interest in this area. Chapter 11 discusses passive bistatic radar in detail, providing a historical perspective; details of bistatic radar geometry and fundamental operation; characteristics of plausible, passive bistatic radar waveforms, such as FM and DTV broadcast, cell-tower emissions, and wireless computer network signals, via the complex ambiguity function; processing requirements; and a survey of some practical systems. Digital modulation has been a boon to passive bistatic radar interest, owing to the potential for reasonably good range resolution. As Chapter 11 details, digital signal processing (DSP) is critical to passive bistatic radar utility, allowing the receive site to broadly capture scattered transmit energy through the formation of multiple surveillance receive beams; enabling direct path receipt and creation of the replica signal needed for pulse compression; allowing pulse compression implementation with different Doppler hypotheses to filter scaled, time-delayed versions of the replica signal; and providing a mechanism to mitigate the impact of the strong, direct path signal interfering with the surveillance channels. Chapter 12 discusses radar application to air traffic control. Air traffic control radar systems are used throughout the world to maintain safe and efficient aviation. These radar systems have a long and proud heritage. While the role of air traffic control radar is evolving due to direct broadcast navigation systems, radar will continue to be pivotal in commercial aviation safety. Chapter 12 looks at the objectives of air traffic control, discusses the purpose of primary and secondary surveillance radar capabilities, and describes design issues for both surveillance modes; this chapter considers requirements for detection of weather effects as well. From its earliest days, it was known that radar detects weather phenomenon. Most people are familiar with radar due to its extensive use on weather newscasts, and the term Doppler radar is widely recognized for this reason. Chapter 13 discusses weather radar in detail, surveying available weather-surveillance radar systems, describing the radar range equation and Doppler processing for weather surveillance, characterizing weather volume reflectivity, and discussing the manifestation of distinct effects (e.g., rainstorm versus tornado) in the weather-surveillance radar product. In addition to weather radar outputs showing up on the evening news, terminal Doppler weather radar detect downbursts and wind shear in support of aviation safety, and aircraft use radar to

13

14

CHAPTER

1

Radar Applications

avoid localized weather. The incorporation of polarization to characterize raindrop size is a current endeavor. Advanced concepts for weather surveillance include the Multifunction Phased Array Radar (MPAR), the newest design from the Federal Aviation Administration (FAA) whose purpose includes replacing aging air traffic control radar and providing a simultaneous capability to monitor weather. During the Vietnam War, insurgents realized they were safely hidden under foliage from the X-band fire-control radars of the time. Shorter wavelengths associated with higher-frequency radar are known to poorly penetrate foliage. The need to detect and engage troops under foliage drove the development of foliage-penetrating radar. Of all the technologies available for surveillance of concealed targets, radar is the most appealing. Chapter 14 discusses the history of FOPEN radar and then focuses on key issues around forming SAR images using lower-frequency, ultrawideband airborne radar. The chapter characterizes propagation through foliage as a function of frequency, examines clutter and target properties, and details SAR image-formation processing. Due to the overlap of FOPEN radar waveforms with a preponderance of other signal sources, radiofrequency interference mitigation techniques are critical in FOPEN; in this regard, Chapter 14 discusses waveform design approaches and both transmit and receive-side waveform notching. FOPEN radar systems leverage polarization to assist in separating manmade and natural objects and in enhancing target characterization, important issues included in this chapter’s exposition. Ground-penetrating radar (GPR) is used to detect buried mines in military applications. GPR is also widely used by commercial industry to detect buried utilities. Moreover, GPR is used in archaeology and has been deployed in emergency management situations to detect life signs under rubble. An extensive discussion on GPR application, principles, and system design is given in Chapter 15. GPRs typically operate at lower frequencies of several MHz, but they can be deployed at operating frequencies in the microwave regime; frequency selection is a function of the properties of the material to penetrate, as well as target features. The system typically couples to the surface via direct contact of the transmit and receive antenna system. Chapter 15 discusses hardware implementation issues and provides sample product outputs. The chapter also more broadly discusses materials-penetrating applications, such as the characterization of objects within concrete building material. The final application considered in this book is police radar. Police radar is used to calculate the speed of roadway traffic. As in the case of weather radar, police radar is well known to the general public. Chapter 16 discusses police radar in significant detail. Current police radar systems are CW (see Chapter 2), operate at X-band, and apply Doppler processing to generate range–rate estimates. These radar systems were initial by-products of radar development during World War II, thereafter leveraging technology readily available at the time to implement product improvement. This chapter also discusses sources of error in police radar application and steps taken to improve deployment.

1.8

COMMENTS

While titled Radar Applications, this book is only able to cover select applications due to the vastness of the radar discipline; in this sense, Select Radar Applications is a more precise title. Important topics are excluded from the text for practicality’s sake, something we certainly regret.

1.9

References

The reader may also notice that some topics are not basic principles, nor are they techniques; they appear closer in alignment to applications. The chapters on CW radar and mmw radar fall into this category. So, an argument can be made that Select Radar Technology and Applications is even more accurate titling. This consternation aside, we hope the reader benefits from the detailed descriptions provided in this book, Radar Applications. The chapters herein build on the legacy of the first two books in the Principles of Modern Radar series; taken as a whole, many important aspects of modern radar principles, techniques, and applications have been covered.

1.9

REFERENCES

[1] M.A. Richards, J.A. Scheer, and W.A. Holm, Editors, Principles of Modern Radar: Basic Principles, SciTech Publishing, Raleigh, NC, 2010. [2] W.L. Melvin and J.A. Scheer, Editors, Principles of Modern Radar: Advanced Techniques, IET/SciTech Publishing, Raleigh, NC, 2012. [3] J.B. McKinney, ‘‘Radar: a case history of an invention,’’ IEEE AES Systems Magazine, Vol. 21, No. 8, Part II, August 2006. [4] M.I. Skolnik, Introduction to Radar Systems, 2nd Ed., McGraw Hill, New York, NY, 1980. [5] N.J. Willis, Bistatic Radar, Technology Service Corporation, Silver Spring, MD, 1995. [6] G.W. Stimson, Introduction to Airborne Radar, 2nd Ed., IET/SciTech Publishing, Edison, NJ, 1998. [7] W.G. Carara, R.S. Goodman, and R.M. Majewski, Spotlight Synthetic Aperture Radar: Signal Processing Algorithms, Artech House, Inc., Norwood, MA, 1995. [8] C.V. Jakowatz, D.E. Wahl, P.H. Eichel, D.C. Ghiglia, and P.A. Thompson, Spotlight-Mode Synthetic Aperture Radar: A Signal Processing Approach, Kluwer Academic Publishers, Boston, 1996. [9] R.J. Mailloux, Phased Array Antenna Handbook, Artech House, Boston, MA, 1994. [10] W.L. Melvin, ‘‘A STAP overview,’’ IEEE AES Systems Magazine – Special Tutorials Issue (Ed. Prof. Peter Willett), Vol. 19, No. 1, January 2004, pp. 19–35. [11] W.L. Melvin, ‘‘Space-time adaptive processing for radar,’’ Elsevier Electronic Reference in Signal, Image and Video Processing, Academic Press, 2013. [12] R.K. Hersey, W.L. Melvin, and E. Culpepper, ‘‘Dismount modeling and detection from small aperture moving radar platforms,’’ in Proceedings 2008 IEEE Radar Conference, May 2008, Rome, Italy. [13] K. Olsen and K. Woodbridge, ‘‘Performance of a multiband passive bistatic radar processing scheme – Part II,’’ IEEE AES Systems Magazine, Vol. 27, No. 11, November 2012, pp. 4–14. [14] M. Cotton et al., ‘‘Developing forward thinking rules and processes to fully exploit spectrum resources: an evaluation of radar spectrum use and management,’’ in Proceedings of ISART 2011, NTIA Special Publication SP-12-485, July 27–30, 2011, Boulder, Colorado. See http://www.its.bldrdoc.gov/publications/2669.aspx. [15] L.B. Fertig, M.J. Baden, J.C. Kerce, and D. Sobota, ‘‘Localization and tracking with multipath exploitation radar,’’ in Proceedings 2012 IEEE Radar Conference, Atlanta, GA, May 7–11, 2012. [16] F.E. Nathanson, Radar Design Principles, 2nd Ed., SciTech Publishing, Inc., New Jersey, 1999.

15

16

CHAPTER

1

Radar Applications

[17] E.F. Knott, J.F. Schaeffer, and M.T. Tuley, Radar Cross Section, 2nd Ed., SciTech Publishing, Raleigh, NC, 2004. [18] D.C. Schleher, Electronic Warfare in the Information Age, Artech House, Boston, Massachusetts, 1999. [19] W.L. Melvin, R. Hancock, M. Rangaswamy, and J. Parker, ‘‘Adaptive distributed radar,’’ in Proceedings 2009 International Radar Conference, Bordeaux, France, October 2009. [20] J.N. Entzminger, C.A. Fowler, and W.J. Kenneally, ‘‘JointSTARS and GMTI: past, present and future,’’ IEEE Transactions on Aerospace and Electronic Systems, Vol. 35, No. 2, April 1999, pp. 748–761.

CHAPTER

Continuous Wave Radar

2

Christopher J. Baker, Ohio State University, Electro Science lab, Department of Electrical and Computer Engineering, and Samuel O. Piper, Georgia Tech Research Institute, Sensors and Electromagnetic Laboratory

Chapter Outline 2.1 2.2 2.3 2.4 2.5 2.6

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . Continuous Wave Radar. . . . . . . . . . . . . . . . Frequency Modulated CW Radar . . . . . . . . . Other CW Radar Waveform Designs . . . . . . FMCW Radar Applications . . . . . . . . . . . . . . References. . . . . . . . . . . . . . . . . . . . . . . . . .

2.1

........ ........ ........ ........ ........ ........

....... ....... ....... ....... ....... .......

........ ........ ........ ........ ........ ........

..... ..... ..... ..... ..... .....

17 21 26 63 67 82

INTRODUCTION

Continuous wave (CW) radar systems continuously transmit an illumination signal and simultaneously continuously receive echo reflections scattered from objects. If an illuminated object is stationary, the frequency of the echo signal is unchanged from that transmitted. However, if an object is moving, then the frequency of the echo signal is altered due to the Doppler effect. By detecting this Doppler frequency, the object’s motion can be determined. The faster the object moves in a given direction, the larger the Doppler frequency. The operation of a CW radar system is shown schematically in Figure 2.1-1. CW radar systems are generally used in compact, short-range, low-cost applications and are often manufactured using solid-state technology. CW radar systems have

f y f t + dp quenc e r F d ve Recei

city v t Velo Targe

FIGURE 2.1-1 ¢ A CW Radar Using the Doppler Shift to Detect a Moving Object.

ngth λ avele W , f ncy t reque itted F m s n a Tr

17

18

CHAPTER

2

Continuous Wave Radar

been used in a wide variety of applications such as the measurement of liquid levels in industrial storage tanks, vehicular speed determination for police speed guns, short-range navigation, missile seekers, fusing, battlefield surveillance, aircraft detection, and automobile cruise control. Indeed, as automobile radar systems are set to become standard for every new car manufactured, CW radar systems may shortly become the most commonplace of any radar variant. The Doppler effect, on which simple CW radar relies, is caused by relative motion. If we consider the signal in Figure 2.1-1, the echoes that return to the radar are effectively compressed by the action of the moving object (or expanded if the object were moving away from the radar). This compression acts to reduce the wavelength of the reflected signals at the receiver, so the observed frequency is increased. The higher the speed of the object, the greater the wavelength compression and therefore an increase in the Doppler frequency shift is observed. The velocity of an object in the radial direction with respect to the radar is related to the Doppler frequency shift, fDop by lfDop (2.1-1) V¼ 2 where V ¼ radial velocity of the object (m s1), l ¼ wavelength of the CW signal (m), and fDop ¼ Doppler frequency (Hz). In other words, the Doppler frequency is scaled by the wavelength of the illuminating signal to convert it to a measurement of velocity, and the factor 2 represents the two-way path traveled in transmission and reception. The simple concept just outlined is known as an unmodulated CW radar system in that a pure tone is used to measure the Doppler shift from a moving object. A drawback of such a CW radar system is that it is unable to detect stationary objects or measure the range to an object (because range is ambiguous to a wavelength). These limitations can be overcome by modulation of the transmitted signal. The modulation imparts a code on the CW signal as a function of time and therefore radial range. For example, the frequency of the transmission can be linearly changed as a function of time. In this way a particular value of frequency represents a particular time delay and hence can be associated with a particular range. In fact, linear FM modulation is probably the most common form used and will be examined in detail in the next section. This type of system is known as frequency modulated continuous wave (FMCW) radar. Figure 2.1-2 illustrates the relationship between frequency and time for this form of modulation. The time delay, Dt, between the transmitting frequency and the frequency of the echo signal, at any instant in time, is proportional to the distance to an object. The time taken for an echo to be received is given by 2R Dt ¼ (2.1-2) c where R ¼ range from the radar to the object (m) and c ¼ velocity of light (m s1). From Figure 2.1-2 we can see that knowledge of the linear rate of change of the transmitted signal and the difference frequency, fd, between the transmitted and received

Introduction

Frequency

2.1

19 FIGURE 2.1-2 ¢ The Relationship between Frequency and Time for a Linear FMCW Radar System.

y nc ue eq r dF itte m s y an nc Tr ue q re dF ive e c Re

∆t fd

∆t =

2×R c

Time

signals allows computation of the time, Dt, for an echo to be received and hence for the range, R, to a detected object to be determined. Figure 2.1-3 shows a schematic illustration of a CW signal composed of a single frequency tone. The amplitude is plotted as a function of the distance traveled by the wave. Thus, the distance between successive peaks is the wavelength, l. If a similar plot were produced but with time on the horizontal scale, then the peak-to-peak length would be the period of the signal and its inverse the signal frequency. The relationship between frequency, f (Hz), and wavelength, l (m), is given by c f ¼ (2.1-3) l where c ¼ velocity of light (m s1). Critical to the understanding of the operation of CW radar is the notion of phase. Phase represents the precise location within a cycle of a signal. It is common to refer phase to the points in time at which the amplitude of the signal becomes zero; it is measured in degrees or radians. A single cycle of a tone maps out 360 degrees or 2p radians of phase. This is equivalent to a vector equal in length to the amplitude of the signal rotating through 360 degrees, that is, making one complete revolution of a circle. By comparing the phase of the outgoing transmit signal with the incoming received signal, fine motion estimation is possible. Figure 2.1-4 illustrates the concept of phase and relative phase. This chapter on CW radar systems focuses on the essential techniques and technologies widely used in today’s systems. The foundations and application of modulated

Amplitude

l

Direction of Wave

FIGURE 2.1-3 ¢ A Portion of a CW Signal.

20

CHAPTER

2

Continuous Wave Radar

FIGURE 2.1-4 ¢ The Top Signal Is Said to Exhibit a Phase of Zero Degrees and the Bottom a Phase of 45 Degrees (with regard to the Top Signal). 45˚

and especially FMCW radar are laid out in the next section. This is followed by a brief discussion of more advanced CW concepts that employ phase and other forms of frequency modulation. Section 2.5 examines the range of applications that exploit the CW technique and summarize a number of the systems used to address these applications. The following is a list of symbols and abbreviations used in this chapter. SYMBOLS AND ABBREVIATIONS ASSOCIATED WITH CW RADAR ACC ADC AGC B Bb c CPI CW DAC dB dBc dBHz dBK dBm ES f F fb fd fDop FFT fm FM FMCW FOV

adaptive cruise control analog-to-digital converter automatic gain control Bandwidth beat frequency bandwidth for point target propagation velocity ¼ 3.0 108 ms1 coherent processing interval continuous wave digital to analog converter decibels decibels relative to the carrier decibels relative to one Hz bandwidth decibels relative to standard temperature decibels relative to a milliWatt electronic support radar frequency noise figure beat frequency difference frequency Doppler frequency shift fast Fourier transform modulation frequency frequency modulation frequency modulated continuous wave field of view

2.2

GHz Gp GSPS k K kHz km LO Lr m MHz ms m/s, ms1 mW ns Pi Pr R RCS RF RPC SNR STC T td Tm tr V VCO VSWR wrt Dfb DF DF0 DR DR0 DT ms

2.2

Continuous Wave Radar

GigaHertz Processing gain Gigasample per second Boltzmann’s constant Kelvin kiloHertz kilometer local oscillator receiver losses meter MegaHertz millisecond meter/second milliWatt nanosecond incident power reflected power range radar cross section radio frequency reflected power canceller signal-to-noise ratio sensitivity time control reference temperature round-trip propagation time modulation period delay between RF and LO paths for internal reflections target velocity relative to the radar voltage-controlled oscillator voltage standing wave ratio with respect to beat frequency resolution frequency deviation, bandwidth of the transmit waveform effective bandwidth of the transmit waveform range resolution ideal range resolution time resolution microsecond

CONTINUOUS WAVE RADAR

CW radars can utilize any part of the RF electromagnetic spectrum just like their pulsed counterparts, and examples exist all the way through from HF to W band. The principal

21

22

CHAPTER

2

Continuous Wave Radar

advantages of CW radars include simplicity, low cost, and small volume – but how do these come about? As CW radars transmit continuously, their peak power and average power are the same. Consider a pulsed radar with a typical pulse length of 1 microsecond and a pulse repetition frequency of 1,000 Hz. A peak power of 1 kW is required to transmit an average power of just 1 W. A transmitter with a peak power of 1 kW makes for a complex and potentially costly system. A CW radar with a peak or average output power of 1 W is straightforward using compact, relatively simple solid-state technology that sometimes costs just a few dollars. A disadvantage of CW radars, though, is reduced dynamic range because of simultaneous transmission and reception. Transmission is continuous and therefore competes with the weak reflected echo signal, which it can easily swamp, thus preventing detection of objects. Of course, the Doppler imparted by a moving target helps mitigate this as the transmission is, in effect, at zero Doppler. This improves the isolation between the transmit and receive signals. To further improve matters, it is usual to use separate antennas for transmission and reception arranged to keep the transmit signal from ‘‘leaking’’ across into the receive antenna. Close-in objects can also cause reflections into the receive antenna, and these again compete with objects detected at longer ranges and also limit sensitivity to slow-moving objects. This usually imparts a limit on the maximum transmit power whereby further increases simply increase the leakage and do not result in an increased detection range. It is largely for this reason that CW radar systems tend to be used in shorter-range applications and consequently tend to be lower in transmission power and smaller in size. There have been one or two exceptions to this and Figure 2.2-1 shows a military 5N62 Square Pair ‘‘Guidance and Illumination Radar’’ capable of a 140–160 NMI range. The larger parabolic antenna is for the transmit signal, and the smaller parabolic section antenna is for reception. Note the dividing blade FIGURE 2.2-1 ¢ Semimobile Configuration of the Improved K-1M Cabin with 5N62 Square Pair FMCW Radar on Display at Kecel in Hungary.

Continuous Wave Radar

2.2

Radar Waveforms CW Radar

Pulse Radar (Amplitude Modulation)

Unmodulated CW

Frequency Modulated CW

Linear FMCW Sawtooth or Triangle

Nonlinear FMCW Sinusoidal, Noise, Pseudorandom

PhaseModulated CW: Biphase, Polyphase

Fixed Frequency

Multiple FrequencyShift Keying

Intrapulse Pulse to Pulse Modulation, Modulation Pulse Frequency Agility Compression Stepped Frequency

Frequency Modulated Linear FM Nonlinear FM

PhaseModulated: Biphase, Polyphase

(termed a knife); it is employed to avoid spillover of the transmit signal into the receive antenna, which would otherwise degrade isolation between transmission and reception and hence reduce detection range. The transmit power is 100 kW CW, and this physical method for improving isolation between the transmission and reception paths is necessary to cope with such high powers. This form of CW radar is much more the exception than the rule, with most being of modest power (usually 1 W or less) and of small physical size. Another feature of some CW radars is that they can operate with a single radio frequency source without the requirement for a separate local oscillator (LO). The relatively low peak transmit power is attractive for solid-state transmit sources and amplifiers that are inherently peak power limited. CW radars also inherently have a lower probability of intercept (LPI) as Electronic Surveillance Measures (ESM) receivers are triggered by the peak power intercepted [1–3]. Figure 2.2-2 shows a segmentation of radar waveforms that are separated into CW and pulsed types. CW and pulsed waveforms are both able to host a wide variety of differing forms of modulation. This wide variety of waveform types, in turn, provides the radar designer with a range of options that allow the performance for a given application to be optimized. CW radar has been in use since World War II and has found widespread application. Perhaps, not unsurprisingly, the applications where CW has found significant and sustained traction are those where low cost, small size, and short range drive system design. Table 2.2-1 lists the most common applications of CW radar system; many have maximum ranges of less than 1 km. As is usually the case with radar, there are notable exceptions such as over-the-horizon (OTH) radars. OTH radars achieve good transmit-to-receive isolation by separating the transmitter and receiver by approximately one hundred kilometers and are thus able to see out to ranges well beyond the horizon.

2.2.1 CW Radar: Configuration Types We briefly introduce the principal CW radar front-end configuration types and discuss their relative merits. Three main forms are examined: (1) autodyne, (2) homodyne, and (3) heterodyne. The simplest of all the configurations is the autodyne. Its configuration uses a freerunning oscillator as the transmission source. Figure 2.2-3 shows the block diagram for

23 FIGURE 2.2-2 ¢ Radar Waveform Hierarchy Showing CW Radar Waveforms.

24

CHAPTER

2

Continuous Wave Radar

TABLE 2.2-1

¢

CW Radar Applications

Radar altimeters Police radars Missile seekers Active protection systems Artillery and missile fuses Doppler navigators Buried object detection Short-range navigation Ship docking Automobile automatic cruise control, collision avoidance Meteorology, weather sensing Tank-level measurement Radar cross section (RCS) measurement range Over-the-horizon radar – long range

FIGURE 2.2-3 ¢ Autodyne CW Radar Block Diagram.

Oscillator

AC Coupling Capacitor Signal Processing

Antenna

Power Supply

the autodyne configuration (sometimes referred to as a self-mixing oscillator or selfoscillating mixer). While stationary targets are not detectable with this CW radar configuration, the return signal from moving targets experiences a Doppler frequency shift. The different transmit and receive frequencies are mixed via the oscillator nonlinearities, yielding a difference frequency equal to the Doppler shift. This difference or beat frequency passes via capacitive coupling to the radar for signal processing. Police speed radar systems generally use the autodyne configuration in the form of a Gunn oscillator source. It is also suitable for intrusion alarms and automatic door opener sensors. However, the simple autodyne CW radar configuration has relatively poor sensitivity, as the noise floor is broadband and the received signal competes with that being transmitted [4] and this limits detection range and velocity estimation. Relative to the autodyne configuration, the homodyne CW radar configuration offers improved sensitivity because of better performance when using a separate mixer and because of the greater transmit-to-receive isolation with separate antennas for transmit and receive. Figure 2.2-4 shows a block diagram of a homodyne CW radar with the transmit signal coupled to the mixer. While a transmit–receive switch cannot be used with CW waveforms, a single-antenna CW radar configuration is feasible (see Figure 2.2-5). This requires a transmit–receive diplexer, such as a circulator, to separate the transmit and receive signals. Different polarizations on transmit and receive have also been used in some CW radars to enhance isolation, although this can add expense.

2.2 Antenna

Continuous Wave Radar

25 FIGURE 2.2-4 ¢ Homodyne CW Radar Block Diagram with Separate Antennas for Transmit and Receive.

Mixer

Signal Processing

Source

Antenna

FIGURE 2.2-5 ¢ Single Antenna Homodyne CW Radar Block Diagram.

Mixer

Circulator

Signal Processing

Source

Downconverter

Receiver Signal Processor

Receive Antenna Low-Noise Amplifier Oscillator

Power Amplifier

Transmit Antenna Upconverter

Transmitter

All approaches have a degree of coupling or leakage that ultimately limits the isolation and hence maximum detection range. Finally, the heterodyne CW radar configuration offers another improvement in performance but at the cost of further complexity [5]. As shown in Figure 2.2-6, the transmit modulation is imparted at a convenient intermediate frequency (IF) and then mixed with the local oscillator to the desired carrier frequency. The receive signal is mixed with the local oscillator, and the signal processing is done at IF. An important advantage of the heterodyne configuration is moving the received signal frequency farther from the transmit frequency because both amplitude modulation and phase modulation noise

FIGURE 2.2-6 ¢ Heterodyne CW Radar Block Diagram.

26

CHAPTER

2

Continuous Wave Radar

power of the transmit signal decreases with separation from the carrier. Filtering improves performance still further by reducing the effects of otherwise broadband noise. The design approach depicted in Figure 2.2-6 offers the potential for good performance with separate transmit and receive antennas enabling good isolation, an RF power amplifier (PA) for higher transmit power, a low noise amplifier (LNA) before the mixer for lower noise figure, and heterodyne mixing for lower amplitude and phase noise. The technology underpinning the design and implementation of CW radar systems is relatively simple and lends itself well to large-volume production in solid-state form. Indeed, the development of FMCW radar technology is resulting in short-range, lowcost systems that have high performance and are able to resolve objects separated by as little as 20 cm in radial range. The main limitation is the ease with which long detection ranges can be achieved. However, there are very many applications in which long range is not a design driver and thus CW systems can be a natural solution. In the next section, linear FMCW modulation is examined in detail, as this is the workhorse of most modern CW systems. FMCW offers a combination of performance and cost that make it highly attractive for many applications, especially for the rapidly emerging area of automotive radar systems, which has enormous commercial potential.

2.3

FREQUENCY MODULATED CW RADAR

In this section, we introduce the linear frequency modulated waveform for CW radar systems. We have already discussed how unmodulated CW radar systems can measure target velocity via the Doppler frequency shift and target direction with a directional antenna; they cannot measure range, however. Measuring range requires a timing reference encoded onto the transmit waveform. In CW systems, this is applied by modulating the frequency or phase of the signal to be transmitted. Although there are other forms of frequency modulation, such as sinusoidal and nonlinear FM, we concentrate on the linear case as it enables the key principles to be introduced and is the most widespread form of modulation in use in CW radar systems today. We also describe the process by which the signal-to-noise ratio and range resolution may be estimated in a linear FMCW radar system together with limits on system performance.

2.3.1 The Linear FMCW Waveform Figure 2.3-1 shows the transmit and receive frequency of an FMCW radar waveform as a function of time. The waveform illustrated is a sawtooth linear FMCW waveform. The receive signal comes from the echo of a target located a distance R from the transmitter. The resulting beat frequency, fb, is the instantaneous difference in frequency between the transmit and receive waveforms. Measurement of the beat frequency allows us to determine the range to a target because it is directly related to target delay. The duration of the linear modulation is set so that it lasts longer than the round-trip transit time for the most distant target to be observed, thus avoiding ambiguities. In Figure 2.3-1 the total peak-to-peak frequency deviation is DF and is termed the modulation bandwidth. The modulation period, Tm, is the time between repetitions of the sawtooth waveform. Together these two quantities (along with their repetition) form a triangle similar but offset to that formed by the beat frequency, fb, and the transit time delay, td (the time difference between the transmit and receive waveforms, that is, the

Frequency Modulated CW Radar

2.3 9.335

2.5

FIGURE 2.3-1 ¢ Transmit-andReceive Frequency as a Function of Time with Beat Frequency Shown.

9.325 2

1.5

9.295

1

Beat Frequency (MHz)

RF Frequency (GHz)

9.315

9.305

9.285 Transmit Frequency Receive Frequency Beat Frequency

9.275

9.265 -

500

1,000

1,500

2,000

2,500

0.5

0 3,000

Time (microseconds)

range to a target). A relation can thus be formed between the modulation bandwidth, the modulation period, the beat frequency, and the transit time that leads to the determination of range to a target: fb DF ¼ Tm td

(2.3-1)

where fb ¼ beat frequency, td ¼ round-trip propagation time delay, DF ¼ modulation bandwidth, and Tm ¼ modulation period. The round-trip propagation time, td, is proportional to range and is given by td ¼

2R c

(2.3-2)

where R ¼ range to target and c ¼ propagation velocity ¼ 3.0 108 m/s. Substituting for td in equation 2.3-1 and rearranging terms leads to the following expression, known as the FMCW equation, which relates beat frequency and range: fb ¼

DF 2R Tm c

27

(2.3-3)

The beat frequency is the product of the frequency sweep slope (i.e., the total frequency deviation divided by the modulation period) and the transit time. Thus, for the

28

CHAPTER

2

Continuous Wave Radar

parameters shown in Figure 2.3-1, assuming a 28-MHz modulation bandwidth and a 1-ms modulation period, the resulting beat frequency is 2.1 MHz, which equates to a target located at a range of 11.1 km. Other formulations of the FMCW radar equation may be constructed. For example, a sawtooth waveform modulation period, Tm, is the inverse of the modulation frequency, fm and thus the FMCW equation can be written as 2R fb ðsawtoothÞ ¼ DF fm (2.3-4) c If the modulation type is a triangular waveform, then there is an additional factor of 2 in the numerator because the period of the triangle wave, including both upsweep and downsweep, is twice that of the sawtooth wave for the same frequency slew rate: DF4R (2.3-5) fb ðtriangleÞ ¼ Tm c The Doppler effect will shift the frequency of the received signals for moving targets, and the beat frequency for an FMCW sawtooth waveform that sweeps up in frequency is given by DF2R 2Vf0 þ (2.3-6) fb ¼ c Tm c Equation 2.3-6 shows the measured beat frequency to be dependent on both Doppler velocity and target range. In other words, moving targets will appear at an incorrect range. The ambiguity resulting from this ‘‘range-Doppler’’ coupling can be resolved if the waveform employs two frequency slew rates or slopes. The triangle waveform with alternate up and down frequency sweeps is a common choice. For the triangle waveform, range is linearly proportional to the difference in the upsweep and downsweep beat frequencies, and velocity is proportional to the sum of the beat frequencies. For the triangle waveform, the upsweep and downsweep beat frequencies are given by fb ðtriangle; upsweepÞ ¼

DF4R 2Vf0 þ c Tm c

(2.3-7)

DF4R 2Vf0 þ Tm c c

(2.3-8)

and fb ðtriangle; downsweepÞ ¼ so that range, R, is given by R¼

Tm c fbðdownsweepÞ fbðupsweepÞ 8DF

(2.3-9)

c fbðdownsweepÞ þ fbðupsweepÞ 4f0

(2.3-10)

and velocity, V, by V¼

Thus, for a single target there are two equations and two unknowns – range and velocity – and hence we can solve for these two quantities. Figure 2.3-2 shows this graphically where separate target range and velocity measurements are made using a triangular modulation waveform. The intersection of the two lines provides the solution to equations 2.3-7 and 2.3-8. The two lines on the graph represent the two different beat frequencies derived from the two different slopes of the triangular modulation and hence

2.3

Frequency Modulated CW Radar

10,000

FIGURE 2.3-2 ¢ Target Range and Velocity Estimation Using Two Different Beat Frequencies Generated by the Two Different Slopes of a Triangular Waveform.

8,000 6,000 –7 MHz/ms slope +7 MHz/ms slope

Velocity (m/s)

4,000 2,000 0 0

5

10

15

20

25

30

35

29

40

45

50

–2,000 –4,000 –6,000 –8,000 –10,000 Range (km)

the actual target range and velocity. In this example, the target range is 10 km and the target was stationary. For multiple targets, an FMCW waveform with multiple pairs of slopes can resolve the range-Doppler ambiguities. The number of FMCW slope pairs that can be supported ultimately determines the number of targets that are resolvable.

2.3.2 Linear FMCW Waveform Trades In practice, choosing FMCW waveform parameters for a given application requires careful consideration of numerous trade-offs. These trade-offs are necessary to ensure that measurements can be made with sufficient accuracy and precision and hence enable determination of target range and velocity within desired limits. The key trade-offs are summarized in Table 2.3-1; they broadly consist of decreasing and increasing the modulation bandwidth, the modulation period, and the beat frequency. We next consider how these parameters influence measurement accuracy. Decreasing the modulation bandwidth reduces the RF bandwidth over which the antenna and other front-end components must operate; generally, this requires simpler components, so costs can be reduced. Decreasing the modulation bandwidth also favors higher output powers and lower amplitude and phase noise. Increasing the modulation bandwidth offers improved range resolution and lower radiated power spectral density. Improved range resolution allows spatial information about targets to be determined, and this can be exploited for modes of operation such as target recognition. Lower radiated power spectral density can reduce the likelihood of the emissions being intercepted because they are spread over a wider frequency range so that the power transmitted per unit frequency is reduced. Decreasing the modulation period will increase both the slope of the modulation and the magnitude of the beat frequency. This helps avoid interference with other forms of modulation that could be applied such as conical scanning demodulation for improved angular location of targets, automatic gain control (AGC), or other internal processing

30

CHAPTER TABLE 2.3-1 Waveform Parameter Modulation bandwidth

Modulation period

Beat frequency

Continuous Wave Radar

2 ¢

FMCW Waveform Trade-offs

Reasons to Decrease

Reasons to Increase

Narrower bandwidth RF components

Finer ideal range resolution

Higher power sources available

Lower amplitude and phase noise

Lower radiated power spectral density

Increased slope

Increased beat frequency

Increased transmit and receive overlap

Increased slope

Increased beat frequency

Avoids interference with conical scan and other modulations

Increased coherent processing interval

Reduced beat frequency chirp or smear due to target motion

Increased effective processed transmit bandwidth and power

Increased unambiguous velocity

Better range and Doppler resolution

Narrower bandwidth linearizer

Lower amplitude and phase noise at greater separation from carrier

Lowered ADC sample frequency

Narrower bandwidth range bin filters

Less beat frequency bandwidth required to cover given range swath

Greater source phase noise correlation

control loops. Increasing the modulation period increases the interval over which processing can be carried out coherently (the coherent-processing interval or CPI). A longer CPI offers the potential for finer receiver frequency resolution and consequently finer range and Doppler resolution. Increasing the modulation period also decreases the modulation frequency and decreases the required bandwidth for the frequency sweep linearity compensation circuitry [6] but at the loss of finer range and Doppler resolution. The FMCW beat frequency signal is estimated via a bank of narrowband filters that form range resolution bins. The beat frequency is usually digitized and the filter bank generated via a fast fourier transform (FFT). This permits convenient application of weighting functions to reduce range side-lobes. The analog-to-digital converter (ADC) and digital processing must have wide dynamic range because radar returns from near and far ranges are present simultaneously. For a 1-ms modulation period, with 500-MHz frequency deviation and 1-km range, the beat frequency is 3.3 MHz, requiring an ADC sampling rate of at least twice that frequency to avoid aliasing. For lower beat frequencies, the receiver filter bandwidth required for each range bin is less for a given range resolution. It also follows that the total receiver bandwidth required to cover a given range swath is less. For higher beat frequencies, a given Doppler shift corresponds to a smaller apparent shift in range. If the linear FMCW radar operates with a constant beat frequency (as might be the case when a limited bandwidth analog frequency filter bank forms the range bins), then the frequency sweep slope must be changed to keep the beat frequencies of interest within the bandwidth of the filters as the range of interest changes. As the transit time decreases for shorter ranges, the slope must increase to keep the beat frequency constant. For constant frequency deviation, this requires shorter modulation periods at shorter

2.3

Frequency Modulated CW Radar

range. Since range resolution is proportional to the product of the range and the ratio of the filter bandwidth to the filter center frequency, the linear FMCW waveform approach has the coarsest range resolution at the longest range.

2.3.3 FMCW Signal-to-Noise Ratio (SNR) Estimation In this section, a methodology is presented for estimating the signal power to noise ratio for both point targets and from area clutter. Example calculations use the PILOT radar as an exemplar. The PILOT radar was originally developed by the Philips Corporation of the Netherlands (now part of the Saab group) as a surveillance sensor for both civil and military applications. The low peak powers of PILOT make it much less likely to be intercepted by an enemy electronic support (ES) system and therefore particularly attractive for short- to medium-range military surveillance. In CW radar, both system noise power and AM and FM noise are important in determining the overall performance, including the computation of SNR. Amplitude and phase noise will be examined in the next section. As with any radar system, the received signal power, Pr, is given by the radar equation: Pr ¼

Pt Gt Gr l2 s ð4pÞ3 ðRÞ4 L

(2.3-11)

where Pt ¼ transmit power, Gt ¼ transmit antenna gain, Gr ¼ receive antenna gain, l ¼ wavelength, s ¼ target radar cross section, R ¼ slant range, and L ¼ loss. The noise power is given by the following equation, N ¼ kT0 BF

(2.3-12)

where k ¼ Boltzmann’s constant ¼ 1.38 10–23 W/(Hz K), T0 ¼ reference temperature ¼ 290 K, B ¼ bandwidth, and F ¼ noise figure. This expression assumes that the receiver components have a physical temperature of 290 K and that the antenna is pointed at an absorbing surface such as the earth. The noise floor of the radar determines the fundamental limit on detection range and is dominated by the performance of the low-noise amplifier if used, or the mixer otherwise. Table 2.3-2 shows a calculation of the noise floor using the PILOT radar parameters. The noise floor allows us to assess the detection performance of the radar by designing the system such that the echo signal is sufficiently larger than the noise so that acceptable values for the probability of detection and false alarm occur. As a rule of thumb, this implies a minimum SNR of 10 dB or greater for reliable detection performance.

31

32

CHAPTER

2

TABLE 2.3-2

¢

Symbol k T0 B F N

PILOT FMCW Radar Noise Power Calculation Description

¼ ¼ ¼ ¼ ¼

TABLE 2.3-3

¢

Symbol Pt Gt Gr l2 s (4p)3 R4 Pr

Continuous Wave Radar

Boltzmann’s constant Reference temperature Bandwidth Noise figure Receiver noise power

Value ¼ ¼ ¼ ¼ ¼

Value

1.38E-20 mW/(Hz K) ¼ –198.6 290 K ¼ 24.6 1,000 Hz ¼ 30.0 3.2 ¼ 5.0 –139.0

(dB) dBm/(Hz K) dBK dBHz dB dB

PILOT FMCW Radar Signal Power Calculation Description

¼ ¼ ¼ ¼ ¼ ¼ ¼

Units

Average transmit power Transmit antenna gain Receive antenna gain Wavelength2 RCS 1984 (Slant range)4 Target signal power

¼ ¼ ¼ ¼ ¼ ¼ ¼

Value

Units

1 1,000 1,000 (0.032)2 (100)2 1984 (44.4)2

W

m2 m2 km4

¼ ¼ ¼ ¼ ¼ ¼ ¼

Value

(dB)

30.0 30.0 30.0 –29.8 20.0 –33.0 –185.9 –138.7

dBm dBi dBi dBsm dBsm dB dBsm dBm

Table 2.3-3 illustrates an example radar range equation computation for the PILOT radar parameters. The table shows that the signal power for a 20-dBsm target RCS at a range of 44 km is 139 dBm. Note that the output power is only 1 W, typical of this class of FMCW surveillance radar systems. An output power of 1 W is less than most mobile phones; we can see how a signal of this strength might be hidden among many other existing emissions in the complex and congested electromagnetic spectrum. Most mobile phones, for example, have an output of between 2 W and 3 W. The antenna gain of 30 dBi and a wavelength of 3.2 cm (corresponding to the 9.3-GHz nominal RF center frequency) are also consistent with radars of this class. In this tabular calculation, the parameter values are converted to decibels, terms in the numerator of equation 2.3-11 are added, and terms in the denominator are subtracted. The maximum transmit power for the PILOT radar is 1 W, but lower power levels are selectable in 10-dB increments down to 1 mW for shorter ranges. The four lower curves in Figure 2.3-3 show the echo signal power for the PILOT radars as a function of range for different selections of output power levels. The noise floor of the radar is also shown sitting at the calculated value of 139 dBm. In addition, the four upper curves show the power levels received by an intercept system as a function of range and PILOT transmitter power. The figures in the box at the top right-hand corner show the maximum radar-detection range and the equivalent intercept range as a function of transmitter power. The intercept ESM is assumed to have a high sensitivity of –80 dBmi. For this example, the detection range of the PILOT radar always exceeds the range at which it can be detected by an ESM system. For this reason, the PILOT radar was often referred to as being ‘‘undetectable.’’ In reality, the probability of detection is much lower than would be the case with an equivalent conventional pulsed radar, but it may still be detected by equipment with sufficient sensitivity. Further, the preceding is a free space computation and does not account for clutter, multipath, and so on. For example, the

2.3

Frequency Modulated CW Radar

33

addition of a forward scatter multipath signal component from a smooth or a rough surface will cause a variation in the range at which an ESM is able to detect the radar. To further illustrate use of the radar range equation across the spectrum of CW radar types, we now examine its application to a level gauge measurement FMCW system. Many materials are kept in tanks used on industrial sites. They might contain noxious substances or for other reasons are not easily accessed. The level-measurement radar provides a means of determining the fill height of the tank without human intervention. Radar makes an ideal measurement tool, and the FMCW technique is used routinely. Figure 2.3-4 shows such a radar system mounted on a tank containing a liquid with a level to be measured.

–40 1 W (28 km) 0.1 W (16 km) 0.01 W (9 km) 0.001 W (5 km) Radar Receiver Noise 1 W (25 km) 0.1 W (8 km) 0.01 W (2.5 km) 0.001 W (0.8 km)

–50 –60

Power (dBm)

–70 –80

FIGURE 2.3-3 ¢ PILOT FMCW Radar Signal Power as a Function of Range along with the Radar Power at an Intercept Receiver.

–90 –100 –110 –120 –130 –140 0

5

10

15

20

25 Range (km)

30

35

40

45

FIGURE 2.3-4 ¢ Liquid Level Measurement Radar Mounted on an Industrial Tank.

34

CHAPTER

2

Continuous Wave Radar

FIGURE 2.3-5 ¢ Modified Bistatic Radar Equation for Vertical Incidence.

Antenna Beam

Range, R

Surface

Apparent Source

Apparent Range, 2R

The detection range may be as little as 1 m; in this case, the time to receive an echo is approximately 6.27 nanoseconds, a demanding measurement task for any radar system. We begin by modifying the radar equation into the bistatic form. This is done so the transmitter can be thought of as being located at its image position due to the reflection from the surface of the liquid as shown in Figure 2.3-5. In effect, this modifies the radar equation into a form closely resembling that of a communications link budget equation. The range measured by the system is therefore twice the actual liquid level height (2R) and the reflectivity coefficient of the liquid (or substance), r, is used rather than area-normalized RCS, s0, to account for the reflectivity from the surface. The radar equation may be written as SNR ¼

Pt Gt Gr l2 ð4pÞ2 ð2RÞ2 LkT0 Bn Fn

ðrÞ

(2.3-13)

where Pt ¼ transmit power, Gt ¼ transmit antenna gain, Gr ¼ receive antenna gain, l ¼ wavelength, r ¼ reflectivity coefficient, R ¼ slant range, L ¼ loss, k ¼ Boltzmann’s constant ¼ 1.38 1023 W/(Hz K), T0 ¼ reference temperature ¼ 290 K, Bn ¼ noise bandwidth, and Fn ¼ noise figure. For CW radars used in level gauge applications, the surface of metals and conductive liquids such as acids and concentrated saline will reflect almost all of the incident power. FMCW radars generally work well for dielectric constants, or relative permittivity, greater than 2. The relative permittivity of water is 80. The relative permittivity of oils is approximately 3.5, and they reflect approximately 10 percent or 10 dB of the incident power. The relative permittivity of hydrocarbons is approximately 1.5, so they reflect 1 percent or 20 dB. The relative permittivity of a vacuum is 1, and the relative permittivity of most gases is approximately 1.

2.3 TABLE 2.3-4

¢

Symbol Pt Gt Gr l2 r (4p)2 (2R)2 L k T0 Bn Fn SNR

SNR Calculation for Level Gauge CW Radar Description

¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼

Frequency Modulated CW Radar

Average transmit power Transmit antenna gain Receive antenna gain Wavelength2 Reflectivity coefficient (4p)2 (2 * Range)2 Loss Boltzmann’s constant Reference temperature Noise bandwidth System noise figure Signal-to-noise ratio

¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼

Value

Units

1 100 100 (0.012)2 0.01 158 (200)2 10 1.38E-20 290 100 1,000 588

¼ 0.0 ¼ 20.0 ¼ 20.0 m2 ¼ –38.3 ¼ –20.0 ¼ 22.0 m2 ¼ 46.0 ¼ 10.0 mW/(Hz K) ¼ –198.6 K ¼ 24.6 Hz ¼ 20.0 ¼ 30.0 ¼ 27.7 mW

Value

(dB) dBm dBi dBi dBsm dB dB dBsm dB dBm/(Hz K) dBK dBHz dB dB

Table 2.3-4 illustrates the SNR calculation for a level gauge at 100-m range with 1-mW transmit power, 20-dBi antenna gain, 0.12-m wavelength, 10-dB loss, 20-dB reflectivity coefficient, 100-Hz bandwidth, and 30-dB noise figure. The parameters are typical of level gauge FMCW measurement radar systems. An increase in power will result in higher levels of signal to noise. However, this does not necessarily translate into improved measurement accuracy as this is usually limited by reverberation caused by reflections from the sides of the tank competing with the reflection from the surface of the liquid. Note also the high value of the system noise figure. This is not atypical of the more simple forms of FMCW radar and is a reflection of performance and cost trade-offs. Sensitivity time control (STC) is a mechanism used to keep the dynamic range of the received echo power matched to the receiver. STC is the programmed variation of the gain or sensitivity of a radar receiver as a function of time within each pulse repetition interval or observation time so that the receiver is not overloaded by strong echoes or by clutter at close ranges. For FMCW, range is proportional to frequency, and the STC function can be realized by high-pass filtering of the beat frequency signal. Since the radar equation tells us that target return power is proportional to range to the fourth power, a 12-dB/octave, or 40-dB/decade high-pass filter, will provide the STC function and hence reduce the receiver dynamic range requirement. Since ADC technology forces a trade-off between faster sample rates and wider dynamic range, this STC filtering enhances the utilization of limited dynamic range ADCs.

2.3.4 Amplitude and Phase Noise For CW radars the amplitude and phase noise sidebands of the transmitter can couple into the receiver and degrade the system noise figure if they exceed the thermal noise level. While AM noise levels are generally much lower than phase modulated noise levels, they may dominate the system noise figure because as described below phase noise correlation effects reduce the effective phase noise level. Stove provides an example in which the AM noise degrades the system noise figure by 3 dB and while the FM noise is 60 dB greater than the AM noise there is a 70 dB phase noise correlation effect which results in negligible phase noise degradation of system noise figure [7]. Phase noise in FMCW radar systems can be significantly higher than thermal noise. This

35

36 FIGURE 2.3-6 ¢ Antenna Reflection and Circulator Leakage Paths in FMCW Homodyne Radar Front End.

CHAPTER

Antenna and Radome

2

Continuous Wave Radar LO Leakage Reflection Mixer Preamplifier Circulator

Coupler Circulator Leakage

RF Power Amplifier

DDS Chirp Generator

RF Oscillator

is due to transmitter to receiver leakage; reflections from close in clutter and noise inherent in components such as voltage-controlled oscillators (VCOs). Phase noise subsequently results in errors in the phase-measuring capability of the radar system. Phase noise is a function of frequency and is usually presented as the ratio of signal power to noise power in a given frequency band that is offset from the transmitter carrier frequency. This section introduces transmitter phase noise effects, including those due to front-end reflections and leakage. Figure 2.3-6 highlights potential antenna reflection and circulator leakage paths in an FMCW radar front end along with the role of the mixer, which also introduces phase noise into the system. The phase noise may be calculated by summing (in logs) all the relevant contributions from the power source through to insertion losses associated with individual components and the appropriate value of phase noise power density at the frequency of operation. Table 2.3-5 shows an example calculation to estimate the phase noise power for the PILOT radar exemplar. This calculation yields a phase noise estimate of 112 dBm due to antenna reflection in a 1-kHz bandwidth at a beat frequency of 1.4 MHz. Significantly, this is more than 27 dB greater than the thermal noise for a 5-dB noise figure. The performance can be improved by providing more isolation between the transmitter and the receiver paths. This is achieved using separate antennas to transit and receive or by introducing a reflected power canceller. For example, with 30-dB additional isolation provided by a reflected power canceller, the phase noise power is reduced to 142 dBm, which is 3 dB less than the thermal noise for a 5-dB noise figure. This assumes a 1-W or 30-dBm source power that is attenuated 0.5 dB by the isolator

2.3 TABLE 2.3-5

¢

Frequency Modulated CW Radar

Antenna Reflection Phase Noise Power Calculation

Parameter

Value

Units

Source power (1 W) Isolator insertion loss Total loss for 20-dB LO coupler Circulator insertion loss Antenna power reflection coefficient – for 2:1 VSWR Circulator insertion loss Phase noise power density at 1.4 MHz Phase noise correlation factor at 1.4 MHz – for 3.3-ns time difference Range bin bandwidth (1 kHz) Antenna reflection phase noise power at 1.4 MHz in 1-kHz bandwidth Reflected power canceller Antenna reflection phase noise power at 1.4 MHz in 1-kHz bandwidth after reflected power canceller

30.0 –0.5 –0.5 –0.5 –9.5 –0.5 –130.0 –30.7 30.0 –112.2 –30.0 –142.2

dBm dB dB dB dB dB dBc/Hz dB dBHz dBm dB dBm

and another 0.5 dB by the coupler that provides the LO drive signal. The circulator insertion loss is 0.5 dB, so a power of 28.5 dBm is presented to the antenna port. Using the same source to transmit and receive a signal reduces the effect of the source phase noise because the noise of the received signal is correlated to that of the transmitted signal [6]. The phase noise correlation factor is termed K2, and it depends on the offset frequency from the carrier and the delay between the transmitted and received signals. For FMCW homodyne radar systems, the carrier offset is the beat frequency, fb. The delay tr is the difference between the source to mixer LO port path and the source to antenna reflection to mixer RF port path. We can write: K 2 ¼ 2 ½1 cos ð2p fb tr Þ

(2.3-14)

1 sin2 a ¼ ð1 cos 2aÞ 2

(2.3-15)

K 2 ¼ 4sin2 ðp fb tr Þ

(2.3-16)

Since

then

or K 2 ffi 4ðp fb tr Þ2 ;

for small products of fb and tr

(2.3-17)

This example assumes that the delay is 3.3 ns, corresponding to a 1-m free space path length difference between the path from the source to the mixer LO port and from the source to the antenna reflection to mixer RF port. This yields a correlation factor of K2 ¼ 30.7 dB. Adjusting the LO path length to cancel the strongest reflected signal path will improve the phase noise correlation and reduce the reflected noise component. This technique will be less effective for applications where antenna scanning results in a changing distance to the reflection point (such as for radome reflections). This static cancellation will also increase the path and decrease the phase noise correlation for other reflected signals and internal leakage signals. Static cancellation will have limited bandwidth.

37

38

CHAPTER

Continuous Wave Radar

2

Lower beat frequencies result in greater phase noise power density and correlation. The phase noise correlation factor is approximately proportional to the square of the beat frequency when the product of fb and tr is small. Over regions where the source phase noise follows a 20-dB/decade slope, these effects will approximately cancel one another, resulting in approximately constant noise power as a function of beat frequency. The phase noise closer to the carrier may typically follow a 30-dB/decade slope, resulting in greater noise for lower beat frequencies. A number of methods have been used to improve phase noise performance, and they generally rely on cancellation techniques that attempt to minimize losses. Figure 2.3-7 shows an example schematic of part of the ELVA FMCW radar system (this is a 200-mW FMCW radar developed for traffic-control applications) in which a phase shifter is inserted between the circulator and antenna to help cancel the antenna reflected power and circulator leakage that otherwise increases the system noise figure and degrades the sensitivity of the radar [8]. Figure 2.3-8 shows the measured power reflected from the antenna along with the power after phase shifter tuning. As an example consider frequencies from 93.8 GHz to 93.9 GHz where the antenna reflection decreases from 26 dB to 29 dB below the transmit level. After phase-shifter cancellation, the power is at least 35 dB below the transmit level. This also highlights that phaseshift cancellation is only effective over a limited bandwidth. For wide bandwidth cancellation, a time-delay method is necessary. Johnson and Brooker [9] describe a reflected-power canceller (RPC) used in a 94-GHz FMCW radar. This manually adjustable vector RPC permits adjustment of the amplitude and the time delay of a signal used to cancel front-end reflections in order to minimize the performance degradation that would otherwise result. The PILOT radar FIGURE 2.3-7 ¢ Block Diagram of ELVA-1 Front End.

FMCW Receiver

P2

Phase Shifter

P1 FMCW Transmitter

200 mW

Antenna ECA-W-10-600

Output Circulator

FIGURE 2.3-8 ¢ ELVA-1 MMW FMCW Radar System Antenna Reflection Power as a Function of Frequency.

93.5 0 –5

93.6

93.7

93.8

93.9

94

94.1

94.2

94.3

94.4

94.5

Antenna Reflection P1+P2 sum

–10 –15 –20 –25 –30

×

×

–35 –40 F, GHz

2.3

Frequency Modulated CW Radar

39

system exploits a dynamic version of RPC as shown in Figure 2.3-9. The RPC uses an X-band vector modulator to adaptively inject a signal with the appropriate amplitude and phase before the mixer to cancel reflected power and thus enhance sensitivity. The RPC offers a 30-dB transmit-to-receive isolation improvement for PILOT, from 20 dB without the RPC to 50 dB with. This compares favorably with the less elegant but pragmatic solution of using separate transmit and receive antennas to achieve sufficient isolation. Table 2.3-6 lists the isolation improvements achieved with four different antennas and signal bandwidths [10]. Mathematically, if a phasor of amplitude A is cancelled using a separate phasor of amplitude, a, and phase error, q, the resulting normalized residual power can be obtained from the law of cosines: r 2 a 2 a ¼1þ 2 cos ðqÞ (2.3-18) A A A It is possible to parameterize the problem using the phase of the residual, allowing the complete set of possible canceling phase and amplitudes to be plotted for any given residual power. In terms of the phase, j, of the residual phasor, the relative amplitude of the canceling phasor is given by a 2 A

r 2 r ¼ 1 2 cos ðjÞ þ A A

Circulator

(2.3-19)

Mixer I Q

Vector Modulator

Control Loop

Transmit Source

TABLE 2.3-6

¢

Reflected Power Canceller Isolation Improvement Performance

Improvement

Bandwidth

Antenna

40 33 40 35

1,200 MHz 400 MHz 650 MHz 1,000 MHz

‘‘Perfect antenna’’ 2.1-m (7-ft) navigation Small navigation 1.5-m (5-ft) navigation

dB dB dB dB

FIGURE 2.3-9 ¢ Reflected Power Canceller Block Diagram.

40

CHAPTER

Continuous Wave Radar

2

The corresponding phase error of the canceling phasor can be obtained from the law of sines: r sin ðqÞ ¼ sin ðjÞ a

(2.3-20)

Figure 2.3-10 shows contours of amplitude and phase errors for cancellation ratio values of 40 dB, 35 dB, and 30 dB. With perfect cancellation amplitude (a ¼ A), a residual power of 30 dB is obtained with q ¼ 1.8 . With perfect phase cancellation, 30 dB is obtained with a relative amplitude of 0.279 dB or þ 0.270 dB. The cancellation phase is constant for a static target or for static clutter and a fixed frequency. However, for linear FM (i.e., a quadratic phase modulation) and a static target, the cancellation phase changes as a function of frequency. If the reflection point is moving, dynamic adaptivity is required. Any lag in the cancellation control loop will lead to a cancellation phase error. An important system characteristic is the loop response time (loop delay), which essentially determines the depth of the cancellation for the FMCW radar. As the frequency sweeps, the delay difference between the feedthrough path and the leakage path causes a time-varying phase difference. For the linear FM case, this will form a frequency difference that is proportional to frequency sweeping rate and the delay difference. This is captured as follows Df ¼

Dfsweep tD Tp

(2.3-21)

where Df ¼ frequency difference between leakage signal and reference signal, Dfsweep ¼ frequency sweep bandwidth, Tp ¼ pulse repetition time, and tD ¼ maximum RF delay mismatch. Relative Amplitude versus Phase Error, Yielding Nulls of –30, –35, –40 dB 0.3 –30 dB 0.2 –35 dB dB of Relative Amplitude

FIGURE 2.3-10 ¢ Contours of Amplitude and Phase Errors for – 40 dB, –35 dB, and –30 dB Cancellation Ratios.

0.1 –40 dB 0

–0.1

–0.2 –0.3 –0.4 –2

–1.5

–1

–0.5 0 0.5 Degrees of Phase Error

1

1.5

2

2.3

Frequency Modulated CW Radar

To compensate this frequency difference, a time-varying error vector has to be generated through the detection of the frequency difference and the extraction of phase information. Because of the finite response time of the loop, the controlling error vector variation always lags behind the phase difference variation, limiting the cancellation performance: Df ¼ 360 Df ts

(2.3-22)

where Df ¼ phase error in degree between leakage signal and feed through signal and ts ¼ maximum loop response time. Substituting equation 2.3-21 into equation 2.3-22, Df ¼ 360

Dfsweep tD ts Tp

(2.3-23)

Lin et al. [11, 12] determined that for a 2-GHz total frequency deviation and a 10-ms modulation period, with a 200-MHz/ms sweeping rate and a 1-ns RF delay mismatch, a 35-dB cancellation requires a phase error of less than 1 . The cancellation loop response time should be less than 13.8 ms, which corresponds to at least 73.5-kHz bandwidth and requires a real-time cancellation control loop. Thus, from equation 2.3-23, we have Df ¼ 360

2 GHz 1 ns 13:6 ms ¼ 360 0:00272 ¼ 0:98 10 ms

For the PILOT FMCW radar parameters in the short-range mode with a 70-MHz frequency deviation and a 1-ms modulation period (and assuming a 1-ns delay mismatch), the cancellation control loop response time must be 40 ms, corresponding to a 25-kHz loop bandwidth. From 2.3-22, we have Df ¼ 360

70 MHz 1 ns 40 ms ¼ 360 0:0028 ¼ 1 1 ms

and in the long-range mode the PILOT frequency deviation is reduced to 7 MHz; with a ten times lower frequency sweep slope, the response time can increase correspondingly by ten times to 400 ms. This equates to a 2.5-kHz loop bandwidth such that the phase error is maintained at 1 ; for example, Df ¼ 360

7 MHz 1 ns 400 ms ¼ 360 0:0028 ¼ 1 1 ms

The phase noise for FMCW radar is a function of frequency. Figure 2.3-11 shows examples of both a simulated and a measured transmitter phase-modulated (PM) noise spectrum for the MMIC-based 77-GHz radar front-end block diagram shown in Figure 2.3-12. Note that the PM noise is approximately 100 dBc/Hz at a 1-MHz offset from the carrier and that the slope is approximately 30 dB/decade. Laloue et al. [13] used a commercially available nonlinear simulator to compute the amplitude-modulated (AM) and PM signal distortion introduced by the nonlinearities of the transmitter on an FMCW signal. They characterized this radar transmitter driven by a sawtooth FMCW

41

42

CHAPTER

2

Continuous Wave Radar PM Noise Spectrum (dBc/Hz)

FIGURE 2.3-11 ¢ Simulated and Measured Transmitter PM Noise Spectrum.

–10 Simulation Measurement

–20 –30 –40 –50 –60 –70 –80 –90 –100

10 kHz 10 kHz T = 25°C 200 M T = 25°C 300 M T = 25°C 400 M Frequency Offset from Carrier (76.5 GHz)

FIGURE 2.3-12 ¢ Block Diagram of an MMIC-based 77GHz Radar Front End.

F_vco=12.75 GHz Chip 1:VCO Multifunction 38.25 GHz MEDIUM Q ×3 RESONATOR (Printed)

1 kHz

RF_out ×2

Chip 3: Transmitter Multifunction F_ref=19 GHz Chip 1: Reference Multifunction HIGH Q ×2 RESONATOR (Dielectric)

RECEIVER

IF2=3*F_vco-2*F-ref

V_tune

RF_in

IF1

LOOP CONTROL MOD

signal issuing from a VCO and successfully compared measured and simulated PM-toAM conversion coefficients. Figure 2.3-13 shows an example of the phase noise spectrum for dielectric resonant oscillators (DROs) that are both free-running (FRDRO) and phase-locked (PLDRO) while operating at 14 GHz. Both have approximately 140 dBc/Hz phase noise at 1 MHz. The FRDRO phase noise increases approximately 20 dB/decade below 1 MHz, while the PLDRO increases approximately 10 dB/decade, with a plateau around 100 kHz [14]. Interruption of the FMCW waveform avoids simultaneous transmit and receive, thus improving sensitivity and allowing longer-range operation. This variation of FMCW – interrupted (IFMCW) – involves turning the transmitter on and off such that the transmission time is matched to the round-trip propagation time and the reception time. This, of course, reduces the transmit duty cycle by a factor of 2, which reduces the average transmit power by at least 3 dB. This waveform approach also offers the benefit of reducing the transmit phase noise effects by more than the loss in average transmit power so that overall the SNR is enhanced (G Brooker [15] page 329).

2.3 –50 1.E+02 –60

1.E+03

1.E+04

Frequency Modulated CW Radar

1.E+05

1.E+06

1.E+07

FRDRO

–70

PLDRO –80 dBc/Hz

–90 –100 –110 –120 –130 –140 –150 Offset Frequency/Hz

Brooker et al. [15, 16] reported on an airborne IFMCW imaging radar that employed a 120-MHz frequency deviation triangle waveform with two fixed 20-ms transmit intervals and 20-ms receive intervals on the upsweep and two on the downsweep. The 20-ms transmit interval results in no eclipsing for targets at the 3-km maximum range. For targets at shorter ranges down to the 1.5-km minimum, the increase in eclipsing loss is more than compensated by the increase in SNR due to the effect of the inverse of range to the fourth power. Almorox-Gonzalez et al. [17, 18] reported an IFMCW radar that staggered the modulation period to help reduce spurious harmonics.

2.3.5 Area Clutter RCS Unwanted reflections from the scattering environment in the form of clutter will dominate radar performance if the echo power of the clutter is significantly above that of system noise. Volumetric clutter, such as that caused by weather, is usually small, and its contribution at short range is rarely significant. However, surface clutter as caused by reflections from the surface of the land or sea and reverberations in a level-measurement radar system can result in echoes significantly above the noise floor. Consider the following example. The RCS of area clutter, sc, is the product of the area-normalized backscatter, s0, and the radar resolution cell area, Ac. If we assume an area-normalized backscatter coefficient, s0 ¼ 30 dB; a range, R ¼ 44 km; an azimuth beamwidth, qaz ¼ 1.3 ; and range resolution, DCR ¼ 24 m, then the clutter RCS, sc, is approximately 17 m2 or 12 dBsm as given by qaz sc ¼ s0 Ac ¼ s0 R pffiffiffi dR 2 0:023 radian ¼ ð0:01Þ ð44; 448 mÞ ð24 mÞ 1:4 ¼ ð0:01Þð713 mÞð24 mÞ ¼ 17:1 m2 ¼ 12:3 dBsm Figure 2.3-14 shows the predicted RCS of sea clutter as a function of range for the PILOT system. The calculations assume a 30-dB sigma zero for the return from the

43 FIGURE 2.3-13 ¢ Phase Noise for an FRDRO and a PLDRO Operating at 14 GHz.

44

CHAPTER

2

Continuous Wave Radar

15

FIGURE 2.3-14 ¢ Sea Clutter RCS as a Function of Range. Clutter RCS (dBsm)

10 5 0 –5 –10 –15 0.0

4.4

8.9

13.3

17.8

22.2 26.7 Range (km)

31.1

35.6

40.0

44.4

sea at shallow grazing angles and a one-way half-power azimuth beamwidth of 1.3 , which corresponds to uniform illumination of a 1.25-m azimuth aperture. At the 44-km maximum range, the RCS of the sea clutter is 12 dBsm, which is just 8 dB less than the 20-dBsm target RCS assumed in Table 2.3-4. In other words, the signal-to-clutter ratio is 8 dB, and targets have to be detected against this background, which is higher than system-induced noise. The clutter may also have non-noiselike statistical properties such as a longer-tailed distribution than is the case for noise, resulting in a further performance degradation.

2.3.6 Range Resolution FMCW radar systems are capable of very-high-range resolution that enables not just highrange measurement accuracy but allows production of high-range resolution profiles (HRRP). This also permits the FMCW technique to be used in a synthetic aperture radar (SAR) mode to generate high-resolution two-dimensional imagery. The range resolution of a linear FMCW homodyne radar is fundamentally limited by the transmit bandwidth. In addition, it will also have limits imposed by transmit-and-receive waveform overlap, receiver frequency resolution, and frequency sweep nonlinearities. For any radar waveform, the range resolution is linearly proportional to the time resolution (i.e., the pulse length for an unmodulated pulsed radar) or inversely proportional to the transmit waveform bandwidth (or the modulation bandwidth for a pulsed radar or an FMCW radar): DR0 ¼

cDT c ¼ 2 2DF

(2.3-24)

where DR0 ¼ ideal range resolution, DT ¼ time resolution, and DF ¼ bandwidth of the transmit waveform. For PILOT, the bandwidths of the transmitted FMCW waveforms are 70 MHz, 28 MHz, and 7 MHz with corresponding ideal range resolution of 2.1 m, 5.4 m, and 21.4 m, respectively.

2.3

Frequency Modulated CW Radar

As seen earlier range measurement is also a function of the beat frequency which must therefore be estimated as accurately as possible. The PILOT radar system uses a sawtooth linear FMCW waveform with 1-kHz modulation frequency and a corresponding 1-ms modulation period. The three range settings of PILOT (4.4 km, 11 km, and 44 km) have corresponding total frequency deviations of 70 MHz, 28 MHz, and 7 MHz, respectively. For ranges out to 4.4 km, with corresponding transit times up to 30 ms, the frequency slew rate slope is 70 MHz/ms that yields a maximum beat frequency of 2.07 MHz. For the 11-km and 44-km maximum range modes, the slope is decreased proportionally so that the maximum beat frequency remains 2.07 MHz. The parameters for PILOT are shown in Table 2.3-7. Figure 2.3-15 shows the beat frequency, fb, spectrum from 0 MHz to 2.048 MHz for the 70-MHz bandwidth and a modulation period of 1 ms. A single stationary point target echo results in a beat frequency at 1.4 MHz, corresponding to its range of 3 km. For example, fb ¼

DF 2R 70 MHz 2ð3 kmÞ ¼ ð70 MHz=msÞð20 msÞ ¼ 1:4 MHz ¼ Tm c 1 ms 3 108 m=s

TABLE 2.3-7

¢

(2.3-25)

PILOT Waveform and Receiver Parameter Summary

Range Setting

4.4

11

44

km

FMCW waveform Frequency deviation, peak to peak Ideal time resolution Ideal range resolution Range resolution claimed Modulation frequency Modulation index Modulation period Frequency slew rate Beat frequency/range ratio Range/beat frequency ratio Maximum transit time Overlap Maximum beat frequency Minimum beat frequency interval Minimum beat frequency spectral width Range resolution limit Analog-to-digital converter sample rate FFT length FFT length FFT frequency sample spacing FFT range sample spacing Window Window frequency resolution (6 dB) Window frequency resolution (6 dB) Window range resolution (6 dB) Doppler frequency shift Range error due to Doppler

Sawtooth 70 14 2.1 2.4 1 70,000 1 70 467 0.002 30 97.0% 2.07 970 1,031 2.2 4.096 4,096 1,000 1,000 2.1 Hamming 1.81 1,810 3.9 62.0 0.13

Sawtooth 28 36 5.4 6.0 1 28,000 1 28 187 0.005 74 92.6% 2.07 926 1,080 5.8 4.096 4,096 1,000 1,000 5.4 Hamming 1.81 1,810 9.7 62.0 0.33

Sawtooth 7 143 21.4 24.0 1 7,000 1 7 47 0.021 296 70.4% 2.07 704 1,421 30.5 4.096 4,096 1,000 1,000 21.4 Hamming 1.81 1,810 38.8 62.0 1.33

Given Given Calculated Calculated Given Given Calculated ms Calculated MHz/ms Calculated Hz/m Calculated m/Hz Calculated ms Calculated Calculated MHz Calculated ms Calculated Hz Calculated m Calculated MHz Assumed Points Given ms Calculated Hz Calculated M Calculated Assumed Sample Calculated Hz Calculated M Calculated Hz/(m/s) Calculated m/(m/s) Calculated MHz ns m m kHz

Given

45

46

CHAPTER

Continuous Wave Radar

2

Figure 2.3-16 is a zoom of the 1.4-MHz single target return showing the beat frequency spectrum from 1.38 MHz to 1.42 MHz (i.e., corresponding to ranges from 2,957 m to 3,043 m). The plot shows sidelobes at approximately 13 dBs that are consistent with a sin x/x or sinc function response for a single-point target. Figure 2.3-17 shows transmit-and-receive frequency as a function of time for two targets. The plot shows the beat frequency as a function of time. For the far-range target,

Beat Frequency Spectrum for Target at 3-km Range

FIGURE 2.3-15 ¢ Beat Frequency Spectrum for Single Stationary Point Target at 3-km Range.

0 Frequency Deviation 70 MHz –5 –10 Magnitude Squared (dB)

Modulation Period 1 ms –15 –20 Range 3,000 m –25 –30 –35 –40 –45 –50

200

400

600

800

1,000 1,200 1,400 1,600 1,800 2,000

Frequency (kHz)

Expanded Beat Frequency Spectrum for Target at 3-km Range 0 Frequency Deviation 70 MHz –5 Modulation Period 1 ms –10 Range 3,000 m Magnitude Squared (dB)

FIGURE 2.3-16 ¢ Expanded Beat Frequency Spectrum for Single Stationary Point Target at 3-km Range.

–15 –20 –25 –30 –35 –40 –45 –50 1,380

1,385

1,390

1,395 1,400 1,405 Frequency (kHz)

1,410

1,415

1,420

2.3

Frequency Modulated CW Radar

Transmit and Receive Frequency

FIGURE 2.3-17 ¢ Homodyne Linear FMCW Overlap. Time

∆F Tm

Beat Frequency fb2 fb1

Time

td1

Beat Freq. Spectra

1/(Tm–td1)

fb2

fb1

Frequency

the beat frequency is larger, as is the transit-time gap after the beginning of the frequency sweep. In the beat frequency spectrum, the nearer-range target has a lower frequency than the far-range target. The spectral width is greater for the longer-range target because the overlap in time is less. The beat frequency bandwidth, Bb, of a single-point target is given by Bb ¼

1 Tm td tsweep recovery

(2.3-26)

where Bb ¼ beat frequency bandwidth of target, Tm ¼ modulation period, td ¼ round-trip propagation time delay, and ts ¼ sweep recovery time. Thus, again, for the PILOT radar with 1-ms modulation period, for a target at a range of 44-km with a 296-ms transit time, and assuming negligible sweep-recovery time, the spectral width of difference frequency is 1,421 kHz, or 42 percent wider than a shortrange target, as shown here: Bb ¼

47

1 1 1 ¼ ¼ 1;421 kHz ¼ Tm td 1;000 ms 296 ms 704 ms

(2.3-27)

Partial overlap between the transmit and receive waveforms causes this broadening of the spectral width of the beat frequency. Figure 2.3-18 shows the beat frequency spectrum for a single point at 30-km range with 7-MHz frequency deviation. The beat frequency is the same as in Figure 2.3-15 because of the ten-time increase in range and transit time is offset by the ten time reduction in reduction in frequency deviation and sweep slope. However, the target return broadens by over 20 percent due to the change in overlap from 98 percent to 80 percent.

48

CHAPTER

Continuous Wave Radar

2

A further source of broadening of the beat frequency is due to target motion causing ambiguity between range and Doppler known as range-Doppler coupling. Figure 2.3-19 shows the beat frequency spectrum for a single-point target at 3-km range with 32-m/s relative radial velocity. The target velocity is sufficient to shift the target return beat frequency by 2 kHz to 1.402 MHz as shown here: fb ¼

DF 2R 2v 7 MHz 2ð30 kmÞ 2ð32 m=sÞ þ þ ¼ ¼ 1, 400 kHz þ 2 kHz Tm c l 1 ms 3 108 m=s 0:032 m

¼ 1:402 MHz Beat Frequency Spectrum for Target at 30-km Range 0 Frequency Deviation 7 MHz –5 Modulation Period 1 ms –10 Range 3,000 m Magnitude Squared (dB)

FIGURE 2.3-18 ¢ Beat Frequency Spectrum for Single Stationary Point Target at 30-km Range with 80 Percent Overlap.

(2.3-28)

–15 –20 –25 –30 –35 –40 –45 –50 1,380

1,385

1,390

1,395 1,400 1,405 Frequency (kHz)

1,410

1,415

1,420

1,415

1,420

Beat Frequency Spectrum for Target at 3-km Range

FIGURE 2.3-19 ¢ Beat Frequency Spectrum for SinglePoint Target at 3-km Range with 32-m/s Velocity.

0 Frequency Deviation 70 MHz –5 Modulation Period 1 ms –10 Magnitude Squared (dB)

Range 3,000 m –15 Velocity 32 m/s –20 –25 –30 –35 –40 –45 –50 1,380

1,385

1,390

1,395

1,400

1,405

Frequency (kHz)

1,410

2.3

Frequency Modulated CW Radar

49

This is also equivalent to shifting the return by two 2-m range bins. This 1.402-MHz beat frequency thus ambiguously also corresponds to a stationary target at 3,004-m range. When the target is moving, there is also a beat frequency chirp during the modulation period due to the relative velocity between the radar and target in addition to the beat frequency components due to range and velocity. If the relative velocity is less than the range resolution divided by the modulation period, then the error will be less than a frequency or range bin. For the preceding moving target example, the beat frequency chirp is only 34 Hz as shown by the following computation: 4DF V DF 2Vt 70 MHz 4Vt Vt ffi 2 ¼ 1 fbchirp ðtÞ ¼ Tm c c Tm c 1 ms 3 108 m=s fbchirp ðt ¼ Tm Þ ¼

70 MHz 4V ð1 msÞ 4ð70 MHzÞ ¼ V ¼ ð0:933 m1 ÞV 1 ms 3 108 m=s ð3 108 m=sÞ

fbchirp ðt ¼ Tm ; V ¼ 32 m=sÞ ¼ ð0:933 m1 ÞV ¼ ð0:933 m1 Þ32 m=s ¼ 34 Hz For the 7-MHz frequency deviation used by the PILOT in the far-range mode, the FM modulation would decrease by 3 Hz. Figure 2.3-20 shows how this results in an increase in the beat frequency sidelobe levels for a case in which the velocity is very high, taking a value of 1.6 km/s. A velocity of this magnitude is so high that the beat frequency is approximately 1.493 kHz higher at the end of the modulation period than at the beginning. Even with a velocity this high, the error only corresponds to approximately 1.5 range bins (note also that sidelobes are only marginally higher). Having examined the spectrum of the beat frequency under various different target conditions, we now consider how well the different beat frequencies can be resolved. This enables the limits on range resolution to be evaluated. The FMCW radar receiver will include frequency filters to form the range bins. The spectral output of the range bin filter will be the convolution of the mixer output beat frequency spectrum and the filter

Beat Frequency Spectrum for Target at 3-km Range

FIGURE 2.3-20 ¢ Beat Frequency Spectrum for Single-Point Target at 2-km Range with 1,600-m/s Velocity.

0 Frequency Deviation 70 MHz –5 Modulation Period 1 ms –10 Magnitude Squared (dB)

Range 3,000 m –15 Velocity 1,600 m/s –20 –25 –30 –35 –40 –45 –50 1,480

1,485

1,490

1,495 1,500 1,505 Frequency (kHz)

1,510

1,515

1,520

50

CHAPTER

2

Continuous Wave Radar

spectral characteristic. The bandwidth of the matched filter will be approximately equal to the inverse of the modulation period less the transit time and any sweep recovery time. Wider bandwidth filters may be used when the frequency sweep nonlinearities broaden the mixer output spectrum beyond the inverse of the modulation period. While analog range bin filters can be used to form the range bins, digital signal processing is attractive for many modern FMCW radar applications. Coarser beat frequency resolution and consequently degraded range resolution will result from using sequences shorter than modulation period. Weighting to reduce frequency and range sidelobes will also broaden the filter bandwidth. The zero-padding technique can be used for narrower bandwidth filters, thus improving both frequency and range resolution. For the ADC at the mixer output to comply with the sampling theorem, the sampling rate fs must be at least two times the maximum beat frequency, or 2fbmax fs

(2.3-29)

where fbmax ¼ maximum beat frequency and fs ¼ sampling frequency. Substituting for fbmax DF 2Rmax 2Vmax þ fs 2 l Tm c

(2.3-30)

Thus, for a modulation bandwidth of 500-MHz frequency and a modulation period of 1 ms, a stationary target at a range of 1 km results in a beat frequency, fbmax, of 3.3 MHz. Thus, fs must be sampled at a rate at least greater than twice this value. 2

DF 2Rmax ¼ 2ð500 MHz=msÞð6:7 msÞ ¼ 2ð3:3 MHzÞ ¼ 6:7 MHz fs Tm c

(2.3-31)

Matching the samples processed in the FFT to the modulation period, and requiring that the number of samples be a power of 2, yields Tm ¼

N 2n ¼ fs fs

(2.3-32)

where N ¼ 2n is the number of samples, so that fs ¼

2n Tm

(2.3-33)

Substituting, 2

DF 2Rmax 2n Tm Tm c

(2.3-34)

4DFRmax 2n c

(2.3-35)

so that

2.3

Frequency Modulated CW Radar

51

So a 213 or 8,192-point FFT is needed for the 500-MHz frequency deviation and 1-km range as shown here: 4DFRmax 4ð500 MHzÞð1 kmÞ ¼ ¼ 6;667 2n ¼ 213 ¼ 8,192 c 3108 m=s

(2.3-36)

This can be accomplished by sampling the 1-ms modulation period at 6.7 MHz and appending 1,525 zeroes to pad the sequence length to 8,192, or increasing the ADC sample rate to 8.192 MHz to get 8,192 samples in 1 ms. Some linear FMCW homodyne FMCW radar systems use coherent-processing intervals that are shorter than the modulation period. This reduces the signal-processing burden since FFTs with fewer points require less processing. Examples include Lear Astronics and TSC [19, 20]. It may also be desirable to only process the interval from the maximum transit time to the end of the modulation period to ensure that any window function applied to the data includes returns from all ranges, avoiding degradation of filter sidelobe response due to misalignment. Figure 2.3-21 shows the beat frequency spectrum for two stationary-point targets with equal RCS located at a nominal range of 3 km with a separation of 4 m. To be able to resolve these as two targets, they have to be resolvable in beat frequency. The null depth between the two target peaks is very sensitive to changes in the range separation that are comparable to fractions of a wavelength because of interference between the range sidelobes of the returns from the two targets. The range sidelobes are approximately 15 dB below the peaks in this figure. Weighting can be used to reduce the sidelobes at the expense of mainlobe broadening; Figure 2.3-22 shows the reduction in range sidelobes to approximately 40 dB below the peak with a Hamming weighting for

Beat Frequency Spectrum for Target at 3-km Range

FIGURE 2.3-21 ¢ Beat Frequency Spectrum for Two Point Targets at 3-km Range with 4-m Separation and Uniform Window.

0 Frequency Deviation 70 MHz –5 Modulation Period 1 ms –10 Magnitude Squared (dB)

Range 3,000 m –15 Separation 4 m –20 Uniform Window –25 –30 –35 –40 –45 –50 1,380

1,385

1,390

1,395 1,400 1,405 Frequency (kHz)

1,410

1,415

1,420

52

CHAPTER

Continuous Wave Radar

2

Beat Frequency Spectrum for Two Target at 3-km Range

FIGURE 2.3-22 ¢ Beat Frequency Spectrum for Two Point Targets at 3-km Range with 4-m Separation and Hamming Window.

0 Frequency Deviation 70 MHz –5 Modulation Period 1 ms –10 Magnitude Squared (dB)

Range 3,000 m –15 Separation 4 m –20 Hamming Window –25 –30 –35 –40 –45 –50 1,380

1,385

1,390

1,395 1,400 1,405 Frequency (kHz)

1,410

1,415

1,420

the same two targets. Note that the depth of the null between the two targets is now much less, and any further reduction could render them unresolved. The achieved range resolution may be estimated as the square root of the sum of the squares of the beat frequency spectral width and the receiver frequency resolution; it is given by vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi !2 u u Tm c t 1 2

DR ¼ þ Dfreceiver 2DF Tm 2R=c tsr

(2.3-37)

where Tm ¼ modulation period, c ¼ propagation velocity, DF ¼ total frequency deviation, R ¼ range, tsr ¼ sweep recovery (or other unprocessed) time, and Dfreceiver ¼ receiver frequency resolution.

2.3.7 Frequency Sweep Nonlinearity Nonlinearities in the frequency sweep result in broadening of the beat frequency spectrum and consequent degradation of the range resolution. Nonlinearities in the frequency sweep can also cause the apparent presence of false targets and the misplacement of real targets, both stationary and moving.

2.3

Frequency Modulated CW Radar

For a VCO, the definition of nonlinearity is the ratio of the deviation from an ideal linear-tuning characteristic relative to the total tuning bandwidth: NonlinearityðVCOÞ ¼

Nonlinearity Bandwidth Frequency Deviation

(2.3-38)

The design aim for linear FMCW radar is for the deviation to be less than the inverse of the modulation period (the modulation frequency) so that the beat frequency spectrum width will be approximately equal to the frequency and range bin bandwidth to avoid degradation of the range resolution. Thus, the nonlinearity criterion may be expressed as Nonlinearity Bandwidth Modulation Frequency

Frequency Deviation Frequency Deviation fm ¼ DF (2.3-39)

NonlinearityðVCO desiredÞ ¼

Linear FMCW radar engineers often also use the following alternative definition for nonlinearity NonlinearityðFMCW Þ ¼

Nonlinearity Bandwidth Nominal Beat Frequency

(2.3-40)

So the FMCW nonlinearity criterion is NonlinearityðFMCW desiredÞ ¼

Nonlinearity Bandwidth Nominal Beat Frequency Modulation Frequency fm ¼ Nominal Beat Frequency fb

(2.3-41)

The alternative definition is useful because it gives the range resolution relative to range. For example, with a 500-MHz total frequency modulation (i.e., a 0.3-m ideal range resolution) and a 1-ms modulation period (a 1-kHz modulation frequency), it is desirable for the variation from an ideal linear FMCW sweep to be less than 1 kHz so that the mixer output spectrum for a point target will be no more than approximately 1-kHz wide. This corresponds to a nonlinearity of 1 kHz relative to the 500-MHz total frequency modulation or 0.0002 percent. For a target at a range of 1 km, the beat frequency is 3.3 MHz. The alternative nonlinearity definition corresponds to a nonlinearity of 1 kHz relative to the 3.3-MHz beat frequency or 0.03 percent. Note that 0.3-m range resolution is 0.03 percent of the 1-km range. Since the frequency deviation is much greater than the beat frequency, the first and more stringent nonlinearity criterion must be used to specify the frequency sweep linearity for the FMCW radar transmitter. The second alternative is useful as an FMCW radar performance metric to represent range resolution relative to the nominal range. Figure 2.3-23 shows the frequency as a function of tuning voltage for a 35-GHz varactor-tuned Gunn diode oscillator. The difference between the actual tuning characteristic and the best linear fit to the characteristic is approximately 40 MHz. This

53

54

CHAPTER

Continuous Wave Radar

2

corresponds to approximately 8 percent nonlinearity and will result in a severely degraded range resolution. Some systems compensate for nonlinearities with piecewise linear fits to segments of the tuning characteristic. Figure 2.3-24 shows the residual errors as a function of tuning voltage for ten segments. This reduces the peak error from approximately 40 MHz to 0.5 MHz near the center of the bandwidth. Compensating for frequency sweep nonlinearities is challenging. Often the output frequency as a function of tuning voltage characteristic is sensitive to temperature so that compensation with an open-loop lookup table or polynomial approach has limited 3.72E+10 3.71E+10 Frequency Deviation Bandwidth

3.70E+10 Frequency (Hz)

FIGURE 2.3-23 ¢ Frequency as a Function of Tuning Voltage for 35-GHz Varactor-Tuned Gunn Diode Oscillator Showing Tuning Nonlinearity.

3.69E+10 Frequency Nonlinearity 3.68E+10 3.67E+10 3.66E+10 3.65E+10 2.00

4.00

7.00 6.00 5.00 Tuning Voltage (V)

8.00

9.00

Ten-Segment Linear Fit Reduces Peak Error from –86 MHz to –2 MHz with Ten-Time Increase in Frequency of Nonlinearity 3.72E+10

1.5E+06 RF Frequency 1.0E+06

3.71E+10

0.0E+00 3.69E+10 –5.0E+05 3.68E+10 –1.0E+06 3.67E+10

Frequency Difference

–1.5E+06

3.66E+10 3.65E+10 2.0

–2.0E+06 –2.5E+06 3.0

4.0

5.0 6.0 7.0 Tuning Voltage (V)

8.0

9.0

Frequency Difference (Hz)

5.0E+05

3.70E+10 Frequency (Hz)

FIGURE 2.3-24 ¢ Ten-Segment Linear Fit Reduces Peak Error from –86 MHz to –2 MHz with a Ten-Time Increase in Frequency of Nonlinearity.

3.00

2.3

Frequency Modulated CW Radar

55

performance. For example, Brooker reported significant performance enhancement with closed-loop linearization relative to open-loop analog compensation [16]. Delay-line discriminators are used to measure the slope in a closed-loop linearizer. The ELVA-1 millimeter wave (MMW) FMCW radar front end uses an active IMPATT frequency multiplier driven by a microwave voltage-controlled master oscillator. The sensitivity and accuracy requirements are met with a triangle waveform generated with an open-loop linearizer to achieve relatively high output power with low noise and good linearity. Table 2.3-8 lists the parameters for the ELVA-1 FMCW radar front end with a 94-GHz RF center frequency. The 100-MHz total frequency deviation corresponds to a 1.5-m ideal range resolution. Figure 2.3-25 shows the beat frequency spectrum for a target at approximately 380-m range. For the triangular linear FMCW waveform with a 100-MHz total frequency deviation, a 5-ms modulation period with 2.5-ms upsweep and a 2.5-ms down sweep, the frequency sweep slope is 40 MHz/ms and the calculated TABLE 2.3-8

¢

ELVA-1 FMCW Radar Front-End Characteristics

Parameter

Value

Units

RF center frequency Total frequency deviation Ideal range resolution, calculated Modulation period, triangle waveform Sweep time, triangle waveform Frequency sweep slope, calculated Range to target Transit time, calculated Calculated beat frequency Measured beat frequency Measured half-power spectral width Measured half-power spectral width Measured half-power spectral width Measured half-power spectral width Measured range resolution Ratio of measured to ideal range resolution

94 100 1.5 5 2.5 40 380 2.53 101,333 102,900 700 350 0.34 0.00035 2.6 1.7

GHz MHz m ms ms MHz/ms m ms Hz Hz Hz Hz Percent of beat frequency Percent of frequency deviation m

Spectrum of the IF Signal Distance between the Target and Radar Is about 380 m 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 100 100 100 101 101 101 101 101 102 102 102 102 102 103 103 103 103 103 104 104 104 Frequency (kHz)

FIGURE 2.3-25 ¢ ELVA-1 94 GHz FMCW Radar Beat Frequency Spectrum.

56

CHAPTER

2

Continuous Wave Radar

beat frequency is approximately 101 kHz for a range of 380 m. In Figure 2.3-25, the peak target return is centered at approximately 103 kHz with approximately 350-Hz halfpower width, which is 0.34 percent of the nominal beat frequency and 0.00035 percent of the frequency deviation. The corresponding range resolution is approximately 2.6 m, which is 1.7 times greater than the 1.5-m ideal range resolution limit. Now we consider a second example based on Alliant Techsystems FMCW radar. Figure 2.3-26 shows the beat frequency spectrum for a single 70-m2 RCS target at 1 km. The half-power spectral width of the target return is approximately 0.3 m. In Figure 2.3-27, the two targets, separated by 0.3 m in range, are resolvable with an approximate 3-dB null between them. The third target is separated by 0.6 m and is even more clearly resolvable. These figures demonstrate a range resolution of approximately 0.3 m, which corresponds to the theoretical limit for 500-MHz total frequency deviation and is indicative of a well-designed linear FM waveform. FIGURE 2.3-26 ¢ Alliant Techsystems FMCW Radar Beat Frequency Spectrum for Single Target.

IF Output Spectrum

Magnitude (2dB/division)

70 m2 Target at 1 km

988

998

IF Output Spectrum

Magnitude (5dB/division)

FIGURE 2.3-27 ¢ Alliant Techsystems FMCW Radar Beat Frequency Spectrum for Three Targets.

Range (m)

995.5

0.3 m

0.6 m

Multiple Targets at 1 km

Range (m)

1,000.5

2.3

Frequency Modulated CW Radar

Consideration of sinusoidal nonlinearities is mathematically tractable. The assumption of a sinusoidal nonlinearity is reasonable for describing the variation from an ideal linear frequency sweep due to inherent VCO tuning nonlinearities. These are the result of residual errors after linearizer compensation or quantization in frequency steps. A simple analysis is presented here on the impact of a sinusoidal nonlinearity on the beat frequency. By assuming that the frequency sweep nonlinearity has a sinusoidal characteristic, the standard deviation of the slope of the frequency sweep can be related to the maximum frequency variation from an ideal linear sweep. The frequency during the down sweep is given by FðtÞ ¼ f0 þ

DF DF t þ A sin ð2pftÞ; 2 T

0 > > t þ tþ ½1 cos 2pft > 2p f0 > < = 2 2Tm 2pf sb ðtÞ ¼ c0 cos > > > > 2p f0 DF ðt td Þ þ DF ðt td Þ2 þ A ½1 cos 2pftðt td Þ > > : ; 2 2Tm 2pf (2.3-51) Simplifying,

DF DF A 2 td ½cos 2pf ðt td Þ cos 2pft ð2td t td Þ þ sb ðtÞ ¼ c0 cos 2p f0 þ 2 2Tm 2pf (2.3-52)

Regrouping, sb ðtÞ ¼ c0 cos 2p

DF DF 2 DF A td ½cos 2pf ðt td Þ cos 2pft f0 t þ td t þ 2 2Tm d Tm 2pf (2.3-53)

2.3

Frequency Modulated CW Radar

Since td ¼ 2R/c, 2 3 DF 2R DF 2R 2 DF 2R þ tþ7 6 f0 2 c 2Tm c Tm c 7 6 sb ðtÞ ¼ c0 cos 2p6 7 5 4 A 2R cos 2pf t cos 2pft 2pf c

(2.3-54)

The beat frequency signal can be written as sb ðtÞ ¼ c0 cos 2pðf þ ft þ eðtÞÞ where DF 2R DF 2R 2 f ¼ f0 2 c 2Tm c

(2.3-55)

2DFR f ¼ Tm c A 2R eðtÞ ¼ cos 2pf t cos 2pft 2pf c For stationary radar and target, the beat frequency signal includes a constant phase term, a constant frequency term, and the nonlinearity term. For constant relative velocity between the radar and the target R ¼ R0 þ Vt, substitute that into the beat frequency signal expression: 2 3 DF 2R0 DF 2V f0 þ þ f0 þ tþ 6 7 c 2 2 c 6 7 6 7 2 6 DF 4R0 þ 8R0 Vt þ 4V 2 t2 DF 2R0 DF 2V 2 7 7 (2.3-56) sb ðtÞ ¼ c0 cos 2p6 t t þ 6 2T 7 Tm c Tm c c2 6 m 7 6 7 4 A 5 2R0 2V cos 2pf t t cos 2pft c 2pf c Combining like terms, 2

3 DF 2R0 2DFR20 þ 6 f0 2 7 c Tm c2 6 7 6 7 6 7 DF 2V 2DFR 2V 0 6 7 1 þ tþ 6 f0 7 6 7 Tm c 2 c c 6 7 sb ðtÞ ¼ c0 cos 2p6 7 V 6 2DFV 7 6 7 1 t2 þ 6 Tm c 7 c 6 7 6 7 4 A 5 2V 2R0 cos 2pft cos 2pf t 1 c 2pf c

(2.3-57)

Equation 2.3-57 shows that there is a constant phase term that is dependent on target range and a Doppler frequency shift term that is linearly proportional to velocity. The constant phase term is approximately equal to the number of wavelengths to the target

59

60

CHAPTER

2

Continuous Wave Radar

and back. For f0 ¼ 35 GHz, DF ¼ 500 MHz, c ¼ 3.0 108 m/s, Tm ¼ 1 ms, and R0 ¼ 1 km then DF 2R0 2DFR20 fb ¼ f0 þ þ ¼ 235,000:0 þ 11:1 ¼ 235,011:1 c Tm c2 2

(2.3-58)

The beat frequency for moving target and sinusoidal nonlinearity is given by 8 9 DF 2V 2DFR0 2V > > > > > 1 > f0 þ þ > > > > T c 2 c c m > > > > > > < = 2DFV V tþ fb ðtÞ ¼ 2 1 (2.3-59) > > Tm c c > > > > > > > > > 2V 2V 2R0 > > > > > sin 2pf t 1 : A sin 2pft 1 ; c c c which can be written as fb ðtÞ ¼ fd þ fr þ fc ðtÞ þ eðtÞ where

DF 2V fd ¼ f 0 þ ¼ 231:7 Hz=ðm=sÞV 2 c 2DFR0 V ffi ð3:3 kHz=mÞR0 fr ¼ 1 c Tm c 2DFV V t ffi ð6:7 kHz=mÞVt 1 fc ðtÞ ¼ 2 Tm c c 2V 2V 2R0 eðtÞ ¼ A sin 2pft 1 sin 2pf t 1 c c c

(2.3-60)

The effect on the beat frequency chirp can be illustrated with the example shown in Table 2.3-9. For f0 ¼ 35 GHz, DF ¼ 500 MHz, c ¼ 3.0 108 m/s, Tm ¼ 1 ms, and TABLE 2.3-9

¢

Comparison of Constant Range and Constant Velocity Examples

Initial range Range at end of modulation period Change in range Velocity Initial Doppler frequency shift Doppler frequency shift at end of modulation period Change in Doppler frequency shift Beat frequency Beat frequency at end of modulation period Change in beat frequency during modulation period

Constant Range

Constant Velocity

Units

1,000.0 1,000.0 0.0 0 0 0

1,000.0 999.7 0.3 300 69,500 70,500

m m m m/s Hz Hz

0 –3,333,333 –3,333,333 0

1,000 3,263,833 3,261,833 2,000

Hz Hz Hz Hz

2.3

Frequency Modulated CW Radar

61

R0 ¼ 1 km, the beat frequency is 3.333 MHz from the start of the target return until the end of the modulation period. For a target approaching with 300-m/s relative velocity, there will be a 0.3-m change in range during the modulation period and a 1-kHz change in Doppler frequency shift due to the change in transmit frequency during the modulation period. Consequently, the beat frequency will decrease from 3.264 MHz initially to 3.262 MHz at the end of the modulation period, which is a beat frequency chirp of approximately 2 kHz. The beat frequency chirp during the frequency sweep induced by the target velocity is approximately twice the modulation frequency in this example. The beat frequency spectrum is very sensitive to frequency sweep nonlinearities [21, 22] and is illustrated in Figure 2.3-28, which shows the beat frequency spectrum for the two targets at 3-km range with 4-m separation. In this case, there is one-half cycle of nonlinearity over the 1-ms modulation period so the nonlinearity frequency is 500 Hz. The peak deviation from an ideal linear sweep is 50 times the 1-kHz frequency resolution, so the nonlinearity amplitude is 50 kHz. This 50-kHz peak nonlinearity is only 0.07 percent of the 70-MHz total frequency deviation. In this example, the two frequencies are resolvable but only just. Figure 2.3-29 shows a different example with ten cycles of nonlinearity over the 1-ms modulation period corresponding to 10-kHz nonlinearity frequency. The peak deviation from an ideal linear sweep is five times the 1-kHz frequency resolution, so the nonlinearity amplitude is 5 kHz. Note that the nonlinearity results in spurious range sidelobes that repeat every 10 kHz in beat frequency, or every 21 m in range. The first sidelobes at 1,400 10 kHz are only 10 dB below the main target return peaks. This 5-kHz peak nonlinearity is only 0.007 percent of the 70-MHz total frequency deviation. This example illustrates the greater sensitivity of the beat frequency spectrum to deviations from ideal linearity when the correlation time of the nonlinearities is shorter. Beat Frequency Spectrum for Two Targets at 3-km Range

FIGURE 2.3-28 ¢ Beat Frequency Spectrum for TwoPoint Targets at 3-km Range with 4-m Separation, Hamming Window, 50-kHz Peak Nonlinearity, and 2-ms Nonlinearity Period.

0 Frequency Deviation 70 MHz –5 Modulation Period 1 ms –10 Magnitude Squared (dB)

Range 3,000 m –15 Separation 4 m –20 Hamming Window –25 Nonlinearity Amplitude 50,000 Hz –30 Nonlinearity Frequency 500 Hz –35 –40 –45 –50 1,380

1,385

1,390

1,395 1,400 1,405 Frequency (kHz)

1,410

1,415

1,420

FIGURE 2.3-29 ¢ Beat Frequency Spectrum for TwoPoint Targets at 3-km Range with 4-m Separation, Hamming Window, 5-kHz Peak Nonlinearity, and 0.1-ms Nonlinearity Period.

CHAPTER

Continuous Wave Radar

2

Beat Frequency Spectrum for Two Targets at 3-km Range 0 Frequency Deviation 70 MHz –5 Modulation Period 1 ms –10 Range 3,000 m Magnitude Squared (dB)

62

–15 Separation 4 m –20 Hamming Window –25 Nonlinearity Amplitude 5,000 Hz –30 Nonlinearity Frequency 10,000 Hz –35 –40 –45 –50 1,380

1,385

1,390

1,395 1,400 1,405 Frequency (kHz)

1,410

1,415

1,420

2.3.8 Direct Digital Chirp Synthesizers An alternative to analog devices is to use direct digital synthesis of the FMCW waveform (although this is not without its own challenges and is the subject of continuing research). One method is to generate the frequency sweep at a lower frequency and then multiply it up to the desired operating frequency. This requires very careful attention to phase noise at the originating frequency because the phase noise increases with the frequency multiplication. For example, Griffiths reported on digitally generating a 40-MHz sweep at 200 MHz, then using two frequency quadrupler stages to yield a 640-MHz sweep at 3.2 GHz [23]. Johnson and Brooker reported on a real-beam FMCW radar that generates a 6-MHz sweep at 7.23 GHz and multiplies up by a factor of 13 to achieve 78-MHz sweep bandwidth at the 94-GHz operating frequency [9]. This approach may be necessary when, for example, the desired frequency deviation is greater than that available from direct digital synthesizer (DDS) technology. Meta et al. present a careful analysis of the impact of nonlinearity on FMCW radar performance along with compensation approaches [24]. Brennan et al. discuss FMCW sweep linearity requirement. Direct frequency synthesizer technology can offer highly linear frequency sweeps. For example, the Analog Devices AD9858 is a DDS that features a 10-bit DAC operating up to 1 GSPS. The AD9858 is capable of generating a frequency-agile analog output sine wave up to 400þ MHz. The AD9858 residual phase noise is 140 dBc/Hz at 100-kHz offset for 403-MHz output. Liu et al. reported on using an Analog Devices AD9858 DDS in an FMCW radar with 800-MHz frequency deviation and 671-Hz modulation frequency. For the corresponding 1.49-ms modulation period, the desired slope is 537 MHz/ms. The AD 9858 frequency update rate is 125 MHz, so there are 8 ns between frequency changes, corresponding to 4.3-kHz frequency steps as shown in Figure 2.3-30. This quantization of the frequency sweep results in frequency errors that

2.4

Other CW Radar Waveform Designs

25,000

FIGURE 2.3-30 ¢ DDS Frequency as a Function of Time.

8 ns (125 MHz) Update – 4.3 kHz Steps 20,000 Frequency (kHz)

800 MHz/1.49 ms Desired 15,000

10,000

5,000

0 0

5

10

15

20 Time (ns)

25

30

35

40

are uniformly pffiffiffiffiffi distributed over 0 to 4,295-Hz, with corresponding 1,240-Hz ð4; 295 Hz= 12Þ standard deviation, which is 185 percent of the 671-Hz modulation frequency [25]. This is a rapidly advancing area in which commercial devices are likely to quickly improve their capability. The Analog Devices AD9912 offers 1 GSPS internal clock speed, up to 400-MHz output directly, integrated 14-bit DAC, and 48-bit frequency tuning word (FTW) so that it can synthesize frequencies in step sizes A , the detection is accepted; otherwise, it is rejected.

&

4

a wide range of pulse-repetition frequency (PRF), generally from low PRF (lookup configuration in air-to-air operations and most of the waveforms used in air-tosurface operations) to MPRF and HPRF (look-down configuration in air-to-air operations); several intrapulse modulations intended for pulse compression (e.g., from no pulse compression in HPRF to large pulse compression factors in LPRF mode – for instance, in imagery modes); and many carrier frequencies over some hundreds of megahertz; in particular, this capability enables (1) the use of frequency diversity to reduce the fluctuations of a target’s RCS and (2) the ability to operate multiple radar systems simultaneously without interference and improved ECCM capabilities.

Powerful programmable signal and data processors, which are key enablers for multimodes and multirole radars.

Finally, a very important element is the radome. It should be as transparent as possible to radar waves without degrading the radiation pattern of the antenna, have a proper aerodynamic shape, be shock resistant (to birds, hail, etc.), and of course protect the radar.

2

The main cause of failure of the transmitter is from the high-voltage generators, not the tube.

4.2

Airborne Fire-Control Radar

FIGURE 4-2 ¢ Left: APG-68 for F-16 [Courtesy Northrop Grumman http://www. northropgrumman.com/MediaResources/Pages/Photo.aspx?pid%3DAN-10134_001%26rel %3D%2F%26name%3DPhotos þ own work (Kemkemian)] Right: RDY for Mirage 2000 [Courtesy Thales Airborne Systems].

Two examples of the latest generation of M-SCAN fire-control radar are shown in Figure 4-2.

4.2.1.1 Air-to-Air Modes Air superiority, air interception, and air combat are the main tasks of the air-to-air mission, which can also include: ●

the air policing mission, which is a peacetime mission and aims to visually identify a doubtful aircraft; and the escort mission, which is the protection of friendly aircraft.

Air superiority is an offensive task that aims to destroy all enemy aircraft. The weaponry used mainly consists of missiles with seekers. The method for long-range airto-air missile guidance will be discussed in the ‘‘Principles of Missile Guidance’’ section. The combat takes place at short distance ( < 10 nm). It uses short-range missiles and even cannon. The typical air-to-air modes of M-SCAN FCR are: ●

velocity search (VS),

range while search (RWS),

track while scan (TWS),

single-target track (STT),

acquisition (ACQ), and

air-combat mode (ACM). Two subsidiary modes mainly triggered in TWS and before firing the weapons are:

raid assessment (RA) and

noncooperative target recognition (NCTR).

In all these tasks, especially in the air-to-air mission, the enemy will try to jam the FCR using electronic attack (EA) (formerly known as electronic countermeasures or ECM) techniques. The specificity of the FCR is to deploy all modes and techniques to

125

126

CHAPTER

FIGURE 4-3 ¢ AI Radar Modes [Jane’s Defence: Air & Space].

4

Fire-Control Radar

Typical AI Radar Modes

Velocity Search

Range While Search

Track While Scan Raid Assessment

mitigate the effects of such EA. These means are called electronic protection techniques (formerly electronic counter-countermeasures). The sequence of these modes in air interception (AI) is shown in Figure 4-3.

Velocity Search The VS mode provides long-range, forward-sector (nose-on), lookup, look-down target detection. The HPRF waveform is used to ensure that the target Doppler frequencies are unambiguously detected, but it causes the target return to be highly ambiguous in range. No ambiguity removal is performed in order to obtain the highest possible sensitivity. Since this method provides only velocity and azimuth information without any range measurement, it is often used cued by an early warning system. Range While Search The RWS mode provides all-aspect (nose-on, tail-on) and allaltitude (look-up, look-down) target detection. RWS provides range and angular data without stopping the normal antenna search pattern. This mode can use different waveforms, depending on the configuration (look-up or look-down) and whether forward-sector target detection or all-aspect (nose-on, tail-on) target detection (see Figure 4-4) is desired using the following waveforms: ●

look-up, all-aspect detection (LPRF);

look-down, focus on nose-on target (HPRF); and

look-down or low-altitude operation, all-aspect target (MPRF waveforms).

The reader is referred to [2] for more details. An automatic waveform management is often used to help the pilot. The reader is referred to Chapter 5 in this volume for a description of methods of ambiguity removal and details on the related waveforms.

Airborne Fire-Control Radar

4.2 90° 60°

120° Doppler Notch

MPRF

30°

MPRF

HPRF

150°

HPRF

0° Fighter

–80

–60

–40

–20

20 Target

40

60

180° 80 NM

Fighter

Target Acquired. Gate Is Large as Motion Has Not Been Determined

Track Is Established. TWS Predicts Position at Next Radar Observation

Track While Scan The TWS mode can track and display multiple target tracks to the pilot. The track-while-scan system maintains the search function, while a data processor performs the tracking functions. The TWS system is capable of tracking many targets simultaneously. Furthermore, the TWS system can also perform a variety of other automated functions such as collision or closest-point-of-approach (CPA) warnings. The TWS system manages targets using gates as depicted in Figure 4-5. A TWS system may use range, angle, Doppler, and elevation gates in order to sort out targets from one another. When a target is first detected, the data processor will assign it an acquisition gate, which has fixed boundaries of range and bearing (angle), and possibly other parameters such as Doppler in coherent modes. When the radar sweeps by the target again, if the return still falls within the acquisition gate, the computer will initiate a track on the target. By following the history of the target positions, the course and speed of the target can be found. The combination of range, bearing, course, and speed at any one time is known as the target’s solution. It is used to predict where the target will be at the next observation. Once a solution has been determined, the computer uses a tracking gate

127 FIGURE 4-4 ¢ HPRF and MPRF Range Comparison [Copy from ‘‘Air and Space borne Radar Systems – Ph. Lacomme, J.Ph Hardange, J.C. Marchais, E. Normant. SCITECH Publishing’’ (Lacomme)].

FIGURE 4-5 ¢ Target Tracking and Acquisition Gates with TWS [SelfDeveloped Graphic (Ballard)].

128

CHAPTER

4

Fire-Control Radar

about its predicted position. If the target falls within the predicted tracking gate, then the computer will refine its solution and continue tracking. Typically, the pilot manually designates the targets to be tracked from the RWS mode. On some modern systems having a large number of tracking estimators, RWS is an embedded part of TWS: The detections automatically trigger tracks. In some multitarget fire-control systems, the tracks on targets to be fired are prioritized, and the scanning domain is automatically adjusted to maintain the tracking on these targets according to the prioritization. When these targets are too widely spread in angle, the MSCAN is no longer capable of maintaining the multiple target tracking and a choice between the targets to track needs to be done by the system. The E-SCAN provides a huge improvement over the M-SCAN on this point.

Tracking Filters The first TWS systems used so-called a–b filters. There are multiaxis (angle, distance, Doppler velocity) recursive filters with filtering coefficients that are predetermined and based on various parameters (age of the track, update rate, error on innovation,3 etc.). The target model used to predict the position of the target at the next scan is a constant velocity model. In the case where the target is turning, the target will probably not be within the tracking gate at the next observation and the tracking system will check to see if the target is within a turning gate (Figure 4-6) that surrounds the tracking gate. (The reader is referred to Volume 1, Chapter 19 for more details.) Given a maximal turn rate, the turning gate encompasses all the area that the target could be in since the last observation. If the target is within the turning gate, the computer starts over to obtain the new target-vector solution. If the target falls outside the turning gate, the track will be lost. The system will continue to predict tracking gates in case the target reappears. Depending on the system, the track file may contain other useful information, such as the classification of the target. This information may be used by the computer when determining the track and turning gates. The operator may also be required to drop the track. More modern TWS systems often use a Kalman with a ‘‘Singer’’ target model. This model states that the residual error component on each axis (range, angle, etc.) is a Turning Gate

FIGURE 4-6 ¢ Use of a Turning Gate to Maintain Track on a Maneuvering Target [Self-Developed Graphic (Ballard)].

Tracking Gate

Predicted Position Observed Position

3

Innovation is the difference between the predicted position at scan k and the target position measured at the same scan k. The innovation is a multidimensional parameter (range error, angle errors, and Doppler velocity error).

4.2

Airborne Fire-Control Radar

129

random process with an exponential autocorrelation function. A Kalman filter continuously computes the covariance matrix of the position estimation, so it adaptively manages the tracking gate and the filtering coefficients provided that the errors’ variances of the ‘‘raw’’ measurements have been correctly estimated. A single target’s model cannot be optimal for all possible target trajectories. This is why, more performing estimators have been developed over the last 15 years. The state of art in tracking systems (for TWS or other tracking modes) relies on the interacting multiple model (IMM) filter or its improved derivatives. Simply speaking, IMM uses several Kalman filters that run in parallel, each using a different model of target motion. Likelihoods with respect to the target path are computed for each model. Using these likelihoods, the IMM forms an optimal weighted sum of the output of all the filters and then rapidly adjust to target maneuvers. Although they were unrealizable before for multitarget-tracking systems because of the computing power requirements, these complex algorithms are now accessible thanks to advances in real-time computing. The description of the IMM algorithm can be found in Volume 1, Chapter 19.

Multitarget Tracking Issues The process of assigning observations with established tracks is known as correlation. During each sweep, the system will attempt to correlate all returns with existing tracks. If the return cannot be correlated, it is assigned a new acquisition gate, and the process begins again. On some occasions, a new target may fall within an existing tracking gate. The system will attempt to determine which return is the existing target and which is the new target, but it may fail to do so correctly. External or not kinematic information can also be used, such as identification of friend or foe (IFF) info if available, the type of target if an NCTR mode is implemented, or data entered by the operator or an external system. Anyway, it is common for TWS systems to have difficulty when there are multiple targets in close proximity or when existing tracks cross each other (Figure 4-7). On recent systems, efficient correlation algorithms are used in dense multitargets situations. The reader is referred to Volume 2, Chapter 15 for more details. On FCR, quick reactions are often mandatory: An important thing for choosing one method rather another is the delay before providing a decision of correlation (thus, a track estimate update) and especially in TWS modes where the refresh rate may be slow. Track Files The TWS system uses a computed track file for each established target it tracks. The track file contains all of the observations that are correlated with that particular target: for example, the range, bearing and time of observation. The track file is given a unique name known as the track designation. This is usually a simple FIGURE 4-7 ¢ Crossing Tracks [Self-Developed Graphic (Ballard)].

Track 1

Overlapping Gates Track 2

130

CHAPTER

4

Fire-Control Radar

number such as ‘‘track 42.’’ Finally, the track file also contains the current solution to the track’s motion. It is a natural extension of the track-while-scan system to create a system that shares tracking information between users: Single radar cannot see the entire tactical situation. All that is required is to transmit the contents of the track file, since it contains all of the observations and the current solution. The sharing of tracking information has been incorporated extensively into modern combat platforms. There are now global command and control networks such as Link 16 that are capable of sharing this information between units all over the world.

Single Target Track In STT, the radar antenna and range gates are continuously kept on the target, keeping a continuous record of the target position. The radar tracks a single target and performs no other function during the time it is tracking. It may be ‘‘cued’’ from another (search) radar, or it may initially find the target in a search mode. The STT mode provides accurate tracking of a single target and can function automatically by engaging auto-acquisition (AACQ) in combat mode or manually by designating a target on the VS, RWS, or TWS displays. Raid Assessment The goal is to get a reliable count of the number of targets within a raid and their relative positions. This task takes place before the targets are engaged. The waveform used has a better range or Doppler resolution than the resolution used, for instance, in the RWS, TWS, or STT modes. The RA is triggered briefly just for the raid analysis from a tracking mode. Noncooperative Target Recognition The ideal goal is to recognize the target type, whether, for example, an F-16, a MiG-29, an Su-30, or a civilian aircraft. In addition to the IFF interrogation, the NCTR has become essential before firing the weapons to avoid fratricide or civilian kills. The current methods rely on high-resolution spectral analysis of the signal from the target. Indeed, the rotating parts of engines – namely, the low-pressure compressor blade – induce periodic modulations on the returned signal. This effect is called jet-engine modulation (JEM).4 It does not identify the target directly, only its engine(s). Similar effects occur with aircraft propellers, ram air turbine (RAT) devices used to power various aircraft pods, and helicopter main and tail rotors; all provide a chopped reflection of the impinging radar signal. The high-resolution spectrum is analyzed, and the result is compared with a library of signatures. A limitation to JEM NCTR techniques, though, is that if the aspect angle is too far from head-on or tail-on and the engine intake or exhaust ducts provide shielding for the jet engine, then there may be no JEM to detect. On the other hand, JEM increases when with an orthogonal orientation to the axis of blade rotation. Another method of NCTR is to make a range profile of the target and compare it with a library of profile (by using a waveform having sufficient range resolution). Acquisition and Air-Combat Mode There are four modes of AACQ with a nominal range of 10 nm (Figure 4-8): ●

4

Wide acquisition (WACQ) displays the target nearest the aircraft after a wide azimuth scan. Typically, the radar scans several horizontal bars (about 30 ) centered on the aircraft’s longitudinal reference axis.

Moving or rotating surfaces on the target will have the same Doppler shift as the target: The modulation is ‘‘carried’’ by the Doppler line of the airframe.

Airborne Fire-Control Radar

4.2

FIGURE 4-8 ¢ Typical Autoacquisition Modes for ACM [Jane’s Defence: Air & Space].

Boresight

Vertical Acquisition

HUD Acquisition

Wide Acquisition

131

In vertical acquisition (VACQ), the radar scans about 50 vertically, centered on the longitudinal axis, and automatically acquires and tracks the first target detected within the specified range. The scan pattern is referenced to own ship axis. This mode is employed when a large altitude separation between the fighter and the target aircraft is anticipated. Heads-up display acquisition (HACQ) displays the target nearest the aircraft after a scan of volume projected off the heads-up (HUD) field of view. This is typically used in the close-in, auto-acquisition, and gun scenario. Boresight acquisition (BACQ) displays the target nearest the aircraft. The HUD shows the antenna beam limits, and the pilot maneuvers to bring a visible target within this circle. This is a very selective acquisition mode since only targets within the antenna beam can be acquired. Once acquired, the target is tracked in STT mode.

4.2.1.2 Air-to-Ground Modes The main missions of an FCR in air-to-ground operation are: ● tactical support, ● ground attack, and ● interdiction. All these missions aim to destroy or neutralize surface targets such as bridges, airfield runways, tank formations, and surface-to-air defense system. These tasks require: ● prior knowledge of the tactical situation provided by other means, ● high-performance navigation systems, ● sophisticated countermeasures, ● adapted weaponry, and ● air protection (escort).

132

CHAPTER

4

Fire-Control Radar

It may be noted that the M-SCAN radars cannot simultaneously perform efficient air-toground tasks and efficient air protection. Thanks to the antenna’s beam agility, the platforms equipped with new E-SCAN radars are now able to carry out air-ground missions while ensuring their own air protection. The main radar modes involved in air-to-surface operation are: ● ●

real-beam ground mapping (RBGP); high-resolution ground mapping, which occurs in Doppler beam sharpening (DBS) and synthetic aperture radar (SAR) modes; update of the inertial navigation system (INS) if the Global Positioning System (GPS) is not available;

ground-moving target indicator (GMTI) and ground-moving target tracking (GMTT);

air-to-ground ranging (AGR) for bombing by using radar telemetry;

bombing using ground mapping; and

assistance to low-altitude navigation through terrain avoidance (TA) and terrain following (TF).

Note that all air-to-surface modes use LPRF waveforms, sometimes with Doppler processing.

Real-Beam Ground Mapping RBGM provides a map of the ground with coarse resolution but over a large area. The quality of the map can be improved on contrasted echoes by using monopulse sharpening. The interest of RBGM is to provide a broad view of the environment. Unlike DBS and SAR modes, it can provide a view along the axis of the velocity vector. Doppler Beam Sharpening and Synthetic Aperture Radar Modes DBS and SAR modes provide high- and very-high-resolution maps of the ground (only on fixed echoes). Both modes use the same principle for the lateral resolution – that is, the movement of the platform for getting a large virtual (synthetic) antenna. ●

The DBS is a simplified version of SAR in which the antenna’s beam sweeps the area to be mapped. The resolutions are on the order of tens of meters in DBS. In the strip SAR mode, the antenna remains steered in a constant azimuth. This mode displays a continuous map along a strip. The resolutions are usually less than a few meters. The length of the synthetic antenna, and thus the resolution, is limited by the duration of the crossing of a target in the main lobe. In the spot SAR, the antenna is slaved to illuminate a small area within the antenna’s beam footprint. This slaving enables very long synthetic antennas. Resolutions as fine as four inches have been successfully achieved.

The reader is referred to Chapter 1 in Volume 1 and Chapters 6, 7, and 8 in Volume 2 for more details. These modes are used for navigation update, detection and accurate localization of fixed targets, and damage assessment after a strike. They may be also used for bombing purposes with GPS- or INS-guided weapons.

Update of the Inertial Navigation System Modern INS are tightly coupled with a GPS unit, making it possible to avoid position and velocity drifts and greatly improving navigation. However, the GPS signal may be unavailable or corrupted due to various

4.2

Airborne Fire-Control Radar

133 FIGURE 4-9 ¢ Top: DBS Map. Bottom: SAR Map [(Courtesy THALES and [2]) Thales Airborne Systems þ own work (Kemkemian)].

reasons. In this case, the update of the INS from radar is automatically obtained in modern systems by correlating radar images with a digital terrain model (e.g., map or terrain profiles). In older systems, the update was obtained thanks to manual designation of landmarks with known coordinates (Figure 4-9).

Ground Moving Target Indicator Ground targets of interest sometime fall into the mobile vehicle category. Their radar signature is at a level that they normally would be masked by ground clutter. If they are moving, though, the difference between the Doppler return of the target and the surrounding clutter can be exploited. In each range gate, a spectral analysis allows separation of the ground clutter area from the thermal noise area where the moving targets can be detected. The detected targets can be accurately localized in azimuth thanks to the monopulse processing. These groundmoving targets can then be tracked (GMTT). Two kinds of GMTI modes are used: ●

Wide area search (WAS) uses a search domain of about 45 in azimuth. The WAS method is easier to use if the detections are superimposed on a digital map. Figure 4-10 shows a typical display, with the concentration of vehicles on main roads clearly visible. In the mode known as SAR/GMTI, GMTI detections are superimposed on a SAR map. Both SAR mode and GMTI mode are performed by the radar so that they appear as a single mode. The SAR mode provides high-resolution images whose size is necessarily reduced. Thus, the SAR/GMTI operates on a small area. A typical display is shown in Figure 4-11. On the aircraft display, the dots indicate the moving targets. In urban, wooded, or hilly areas, the number of ground-moving targets is not visible.

134 FIGURE 4-10 ¢ Typical WAS GMTI Display [(Courtesy THALES) Thales Airborne Systems þ own work (Kemkemian)].

CHAPTER

4

Fire-Control Radar

MTI A L0

40

23

06

09

12

FIGURE 4-11 ¢ Typical SAR/GMTI Display [(Courtesy THALES) Thales Airborne Systems þ own work (Kemkemian)].

Figure 4-12 illustrates the use of both types of GMTI. On the aircraft display, the dots are the moving vehicles.

Detection of Slow-Moving Targets When viewed by a moving radar platform, fixed targets on the ground lie within a particular Doppler bandwidth. Expressed in terms of radial velocity, the Doppler extent of these fixed targets is DVR ¼ VP sinq Dqclut where q ¼ angle of fixed target relative to the aircraft velocity vector, Dqclut ¼ azimuth beamwidth that returns a clutter level exceeding the thermal noise level, VP ¼ velocity of the platform, and DVR ¼ change in radial velocity.

4.2

Airborne Fire-Control Radar

135 FIGURE 4-12 ¢ ‘‘WAS’’ GMTI Superimposed on Digital Map and SAR/GMTI [(Courtesy THALES) Thales Airborne Systems þ own work (Kemkemian)].

“SAR/GMTI”

“WAS” GMTI

Generally, Dqclut % 20–3dB. These considerations show that detection of slow targets is difficult, even impossible, for directions far from the direction of the platform’s velocity vector. A better-performing approach can suppress fixed ground clutter that interferes with the moving-target returns by employing an antenna system divided into multiple subapertures and thus multiple phase centers. These multiple subapertures act as independent antennas to form a so-called displaced phase-center antenna (DPCA). The basic concept is to keep pairs of phase centers motionless from pulse to pulse, simulating an antenna that stays motionless in space. This has the effect of driving the Doppler bandwidth of clutter to zero so it can be canceled on subtraction of the data from these pulse pairs. With the background ‘‘removed,’’ all that remains are moving objects and noise. Moving targets, however, will suffer some amount of loss on subtraction, depending on their range rate and the difference in time between observations. Still, after detection, the location in azimuth of the target remains unknown. Multiple phase centers along the horizontal axis can solve this problem, in an approximate sense, by means of monopulse techniques. With a minimum of three phase centers, both the DPCA technique of clutter cancellation and monopulse techniques for location of targets can be combined to detect and locate targets on the ground. Surveillance radars designed for GMTI typically implement the DPCA technology and have three or more antenna phase centers in azimuth. An adaptive version of DPCA – STAP – is most often used. The reader is referred to Volume 2, Chapter 10 for more details. Such an antenna’s arrangement is difficult to implement on existing multirole FCR (air-to-air and air-to-surface) with M-SCAN antenna. However, this technique will be commonly used on future AESA radars with multiple subarrays.

136

CHAPTER

4

Fire-Control Radar

Air-to-Ground Ranging AGR provides the fire-control computer (FCC) the slant range of ground along the antenna’s beam axis – that is to say, the range where the monopulse ecartometry in elevation is null. The AGR is used for bombing in two manners: ●

FIGURE 4-13 ¢ CCIP Heads-Up Display [Annotated Simulation Screen Capture. Lock-on Modern Air Combat http://lockon.co.uk/ en/modern_air_ combat/].

In the first mode, constantly computed impact point (CCIP), the air-to-ground ranging function of the radar measures the area in the vicinity of the CCIP HUD symbol so the FCC can interpret the correct range and elevation of the target. The impact point is indicated via HUD symbology (Figure 4-13). The FCR can provide ranging information to the target area, allowing impact point symbology to be displayed to the pilot on the HUD. In the CCIP mode, the pilot visually acquires the target but uses HUD CCIP information to determine when to release the designated weapon. The CCIP delivery is appropriate for strafing, forward-firing rocket (FFR), or freefall bomb deliveries. The CCIP delivery is not an all-weather delivery even though radar information is used to assist in the fire-control solution because target acquisition and designation are done visually. In the second mode, constantly computed release point (CCRP), the primary ‘‘full system’’ delivery of unguided and many guided weapons is summarized in the CCRP attack. Delivery parameters vary, depending on the tactical situation. Examples include (a) low-altitude, unaccelerated (1-G), delivery of high-drag, general-purpose (GP) weapons; (b) low-altitude, accelerated ( > 1 G), toss or dive toss of GP and precision-guided weapons; and (c) medium- to high-altitude level or dive or dive toss maneuvers. The onboard FCR gives the option of all-weather day or night delivery. The radar provides range, bearing, and depression angle information to the FCC. Steering information presented to the pilot via the HUD or appropriate multifunction display (MFD) symbology includes prescribed heading and pull-up information for accelerated maneuvers. In general, when the downrange travel for the selected weapon is equal to the radar slant range, a release pulse

4.2

Airborne Fire-Control Radar

CCRP Radar Delivery Depression Angle q

e ang nt R a l ar S Rad

Altitude above Target

137 FIGURE 4-14 ¢ CCRP Radar Delivery [SelfDeveloped Graphic (Ballard)].

Down-Range Travel

FIGURE 4-15 ¢ Target Designation in Geodesic Coordinates with SAR [(Courtesy THALES) Thales Airborne Systems þ own work (Kemkemian)].

is generated to the weapon’s store (Figure 4-14). For a precision-guided weapon such as a laser-guided bomb, radar information is used to provide aircrews with weapons envelope information to allow selection of release parameters. Derived altitude information from the radar elevation array can also be used to compute minimum safe altitudes (blast avoidance) or minimum time-of-flight (TOF) information to allow for appropriate weapon arming.

Bombing Using Ground Mapping These modes are used with a weapon utilizing precise target location data based on coordinates defined by the World Geodetic System of 1984 (WGS 84). ●

INS- and GPS-guided weapons: Precision standoff weapons such as the GBU-31 joint direct attack munition (JDAM), the AGM-154 joint standoff weapon (JSOW), or the French AASM, which all use accurate target designation, require high-resolution SAR modes for precision target identification and location (Figure 4-15). Radar-offset aim point: In some cases, the target level is not sufficient with respect to surrounding ground clutter and thus cannot be localized on the radar map. In fact, the target (e.g., underground bunker, maritime mine splash point) may be completely obscured. Often, though, unique radar-identifiable offset aim points are in the target vicinity, enabling accurate weapon delivery. Mission planning for radar-offset, aimpoint delivery requires a radar-significant aim-point selection along the desired attack heading that is beyond the designated target. This allows the radar offset to be tracked through weapon release. Precise range, bearing, and differential altitude between the aim point and the target are calculated for entry into the fire-control computer. For low-altitude deliveries, consideration should first be given to cultural features with significant vertical development (tall buildings) for enhanced reflectivity and minimum masking. Medium to high-altitude attacks require radar offsets

138

CHAPTER

4

Fire-Control Radar

FIGURE 4-16 ¢ Offset Aim-Point Bombing [SelfDeveloped Graphic (Ballard)].

far enough from the target to not be masked by limited depression angles of the firecontrol radar at release. If EO or IR sensors are also available, the offset aim-point attack may provide initial cueing information to the target area with final target designation and weapon release being accomplished through the sensor fusion process in the fire-control system. Figure 4-16 depicts a typical offset scenario with a radar-significant pier on a river serving as the offset. The typical B-scan radar scope used by the pilot or weapon system operator is depicted in the inset. ●

Radar forward air controller beacon (RABFAC): Offset aim-point radar bombing can be used in close air-support (CAS) missions by use of a ground-based, radar beacon forward air controller (RADFAC) AN/PPN-18, which allows the identification of targets that are not radar significant to the supporting aircraft by a ground forward air controller (FAC). The FAC provides the offset range, bearing, and differential altitude info to the aircrew. The ground beacon provides the radar-significant offset on the radar ground-mapping display (Figure 4-17).

Terrain Avoidance and Terrain Following (TA and TF) Terrain avoidance is a mode in which the aircraft radar continuously sweeps the ground area directly in front of the aircraft in order to avoid mountains. This is particularly useful when clouds, haze, or darkness obscure visibility. Two horizontal bars are generally used, with a maximum range of 10 nm. Below 3 nm, the ground generally is no longer visible, but previously detected echoes are stored and displayed according to the aircraft’s progression. Such a mode is used at low altitude (typically 500 to 2,000 feet). Indeed, in enemy territory, safety is increased when the plane flies at low altitude to take benefit of masks formed by land relief. The pilot chooses the clearance altitude h1 and then must navigate so that the ground echoes are located between h1 and h2 and appear in green on the radar

4.2

Airborne Fire-Control Radar

139 FIGURE 4-17 ¢ RABFAC Beacon CAS [SelfDeveloped Graphic (Ballard)].

1

h1 h–2 Danger Zone Monitored Zone

3

Masked Zone

v

2

Radar Display

display. Echoes above the clearance level appear in red (danger zone). Two types of TA are utilized: 1. contour mapping stabilized in the horizontal plane (Figure 4-18, 1) and 2. contour mapping slaved to the aircraft’s velocity vector (Figure 4-18, 3); this allows perfectly safe blind penetration. A typical radar display is shown in Figure 4-18, 2. Low-altitude navigation ( >500 ft) cannot be used to penetrate safely far into enemy territory: The aircraft should fly as low as possible between 200 and 300 ft. Very-lowaltitude navigation requires an automatic TF system. Such a mode is very demanding as

FIGURE 4-18 ¢ Terrain Avoidance [(from [2]) Manuscript of ‘‘Air and Space borne Radar Systems – Ph. Lacomme, J.Ph Hardange, J.C. Marchais, E. Normant. SCITECH Publishing’’ (Lacomme with permission)].

140

CHAPTER

4

Fire-Control Radar Desired Push-over Point

FIGURE 4-19 ¢ Terrain-Following Profile [SelfDeveloped Graphic (Ballard)].

Desired Pull-up Point

FIGURE 4-20 ¢ Simultaneous Terrain Avoidance and Terrain Following for Very-LowAltitude Navigation [(from [2]) Manuscript of ‘‘Air and Space borne Radar Systems – Ph. Lacomme, J.Ph Hardange, J.C. Marchais, E. Normant. SCITECH Publishing’’ (Lacomme with permission)].

not only all ‘‘normal’’ ground echoes must be detected but also particular vertically extended echoes such as tops of pylons and cliffs. Ideally, such a mode should work within rain without detecting it as obstacle. It should also detect power line cables, among other obstacles. To carry out this task, a vertical scanning mode is used (about 15 ) with a few vertical bars since the revisit rate should be of the order of 1 second to provide sufficient reactivity. TF alone only allows navigation at very low altitude in the vertical plane (Figure 4-19). Simultaneous TA or TF is difficult – even unachievable – with M-SCAN because a wide angular area must be swept in a limited time. E-SCAN overcomes these scanning issues and allows the full simultaneity of TA and TF, enabling a very-low-altitude navigation both in vertical plane and horizontal plane (Figure 4-20).

Precision Velocity Update and Doppler Navigation In this mode, the radar again tracks ground features using Doppler techniques to precisely predict aircraft ground speed and direction of motion. Wind influences are taken into account so that the radar can also be used to update the aircraft’s inertial navigation system when the GPS signal is unavailable or corrupted. 4.2.1.3 Air-to-Sea Modes There are three primary air-to-sea modes: ●

Sea 1 is designed to detect and track at long range large vessels (large RCS) whether fixed or mobile. In a low sea state, it can also detect small targets at shorter range. An LPRF is used without Doppler processing. A target shall be detected if it is sufficiently contrasted with respect to the surrounding clutter, which increases with the sea state. Because the sea clutter is not a Gaussian noise but has a long-tailed probability density function, a special CFAR is required. Sea 2 is designed to detect and track at medium range moving vessels by high sea states. An LPRF is used with Doppler processing. Small targets are detected if they are sufficiently moving to be located on the thermal noise region after Doppler processing.

4.2

Airborne Fire-Control Radar

141 FIGURE 4-21 ¢ ISAR Display [US Navy http://www.nrl. navy.mil/research/ nrl-review/2002/ electronicselectromagnetics/ lipps/].

Inverse synthetic aperture radar (ISAR) is a well-established technique to identify the reflectivity centers of the target with high spatial resolution. It jointly uses highrange resolution profiling (HRRP) and high-resolution Doppler analysis of the received signal. Indeed, a vessel has periodic motions around its center of gravity (roll, pitch, and yaw). Simply speaking, these motions create a Doppler effect proportional to the height of a given reflectivity center: FD ¼

2wh sin j l

where FD is the instant Doppler frequency, w is the instantaneous rotation rate, j is the angle between the line of sight and the rotation vector, and h is the height of the given reflector. A fine two-dimensional reflectivity map (range – Doppler) of the target is generated. After having estimated the instantaneous rotation rate w, the range – Doppler map can be converted in a display such as Figure 4-21. ISAR is normally used to classify and designate targets to be attacked with antiship missiles such as the AGM-84C. Target information is relayed to the missile via the MILSTD 1760D interface and the missile acquires the target with its self-contained active radar missile-control system after launch.

4.2.2 Radar Displays The information issued from the radar can be displayed on both MFD and HUD.

4.2.2.1 Multifunction Displays The multifunction displays are CRT-based displays that sit in the front of the cockpit (Figure 4-22). They are used primarily to display weapons system management functions, in particular, weapons stores information, radar information, and weapons electro-optical displays. The MFDs are also used to display HUD, radar warning receiver (RWR), and navigation information. The MFDs provide an integrated method of accessing the data required to perform a mission. Around the screen there are several control buttons. They are programmed to perform specific single or multiple functions for each MFD format. Each function is identified by a mnemonic displayed adjacent to the control buttons.

142 FIGURE 4-22 ¢ Top: Typical Air-toAir MFD Display. Bottom: Air-toGround Display in DBS Mode [F-16 Block 52 Fighter Lead-In School Manual V 3.0].

CHAPTER

Fire-Control Radar

4

Bug Aspect Angle

Bug Magnetic Ground Track Mode Submode FOV

Control STBY Page Override Access

Sensitivity Indicator

Bug Calibrated Airspeed Bug Closure Rate

Acquisition Cursor

Min/Max Search Altitudes

Range Increment Range Scale Range Decrement Antenna Elevation Tic

Range Marks

Azimuth Scan Width Steerpoint Symbol

Secondary Target (System Track)

Elevation Bar Scan Bull’s-eye Bearing and Range to Cursor Relative Bearing and Range to Bull’s-eye

Azimuth Scan Limit Lines

Antenna Azimuth Tic

4.2.2.2 Heads-Up Displays Regarding information coming from the radar, the typical HUD shows the tracked target, velocity vector, maximum and minimum range, missile launch limits, an allowable steering error circle, and an aim dot. A typical display for STT is shown in Figure 4-23. This mode is highly recommended to reduce distraction as the pilot closes to optimal medium-range missile range.

4.2.3 Weapon Modes 4.2.3.1 Air-to-Air Missile Modes Long ago, the weapons and sensors were short range, to the role of fire-control radar was mainly to guide cannon fire (dogfight). Although the cannon mode is still used as a last

4.2

Airborne Fire-Control Radar

143 FIGURE 4-23 ¢ Semi-active Missile Symbology with STT [Annotated Simulation Screen Capture. Lock-on Modern Air Combat http://lockon.co.uk/ en/modern_air_ combat/].

resort, most weaponry now consists of air-interception missiles (AIMs), which enables strikes from a distance. Currently, the following two types of missiles are in use. 1. Missiles with electromagnetic seekers (EM or radar missile) are intended for beyondvisual-range (BVR) strikes. Current inventory includes two basic types: those with semi-active guidance such as the AIM-7 Sparrow (the target must be continuously illuminated by the FCR because the missile seeker is a receive-only system without a transmitter) and

those with active and autonomous guidance such as the advanced medium-range air-to-air missile (AMRAAM); in fact, the seeker is a small radar.

2. Missiles with passive infrared seeker are used for short to medium distances. The missiles with autonomous seekers – that is, those that do not require illumination from the FCR – allow the engagement of multiple targets (in TWS mode or in search while track, or SWT, mode with E-SCAN). It can, however, also be fired in STT mode or in visual mode. In contrast, semi-active missiles only allow a mono-target fire control. Indeed, the radar must be locked in continuous STT mode or must be in flood mode. In the case of an active seeker, the SNR at reception is proportional to SNR /

ERPSeeker ASeeker sM R42

where ERP ¼ equivalent radiated power (Pe.Ge) by the seeker’s transmitter, ASeeker ¼ antenna’s seeker surface, sM ¼ monostatic RCS of the target, and R2 ¼ distance between the target and the seeker.

144

CHAPTER

4

Fire-Control Radar

In the case of a semi-active missile, the FCR and the seeker form a bi-static radar. The signal-to-noise ratio at the seeker’s receiver level is now proportional to SNR /

ERPFCR ASeeker sB R21 R22

where ERP ¼ equivalent radiated power from the FCR, R1 ¼ distance between the FCR and the target, ASeeker ¼ antenna’s seeker surface, sB ¼ bi-static RCS of the target, and R2 ¼ distance between the target and the seeker. For the sake of simplicity, suppose the reception sections have the same characteristics in both cases and sM ¼ sB. At launch, R1 R2 R, but the ERP of an FCR is much higher than those of an active seeker FCR. All things being equal, at minimum SNR allowing correct tracking: Rsemiactive ERPFCR 0:25 Ractive ERPSeeker The ratio ERPFCR/ERPSeeker is about 10 to 20 dB, depending on the systems. So the tracking range ratio is about 2 to 3. At first glance, one might think that the semi-active guidance offers a longer engagement range. In fact, this analysis is not true: As the active radar seeker has a short range, an active missile such as AMRAAM performs most of its flight in inertial navigation mode guided by the tracking data coming from the FCR and using a particular data link. When it is close enough to the target, it locks its seeker and becomes active and autonomous (fire-and-forget concept). On the one hand, the engagement envelope of a semi-active missile is limited by both: ●

he tracking range of the bi-static radar range formed by the FCR and the seeker and

the propulsion system of the missile and its aerodynamic features.

On the other hand, the engagement envelope of an active missile depends mainly on the missile’s propulsion system and aerodynamic features. Nevertheless, the longer the active seeker range, the more ‘‘fire and forget’’ the missile is, so the less vulnerable the launcher is to an enemy riposte. Figure 4-24 shows the relative envelopes of active and semi-active EM missiles. Three areas are displayed for the active missile: 1. The seeker-only area is the engagement envelope when the missile is fired with its seeker locked (no need of external guidance). 2. The aim point þ seeker area is the engagement envelope when the missile is fired with a predicted target trajectory without data update during the inertial phase. 3. The FCR tracking þ seeker is the usual engagement mode of an active missile: inertial navigation mode guided by the tracking data coming from the FCR, then final guidance by the active seeker.

Airborne Fire-Control Radar

4.2 Engagement Envelope Semi-active Nose-on

FIGURE 4-24 ¢ Relative Comparison of Active vs. Semiactive Missiles Envelopes [http:// www.fas.org/man/ dod-101/sys/ missile/aim-120. htm þ own work (Kemkemian)].

Active

FCR Tracking + Seeker

Semi Active

145

Aim Point + Seeker Seeker Only

Tail-on Aircraft Support Needed Aircraft Support Not Needed

Range NM Altitude Feet (MSL)

30

20

20

10

10 0 10 20

FIGURE 4-25 ¢ Air-to-Air Missile Envelope [Jane’s Defence: Air & Space].

20

10

10

20

30 Range NM

- Fighter - Target Typical Missile Envelope against a Co-altitude, Non-maneuvering Target

Another example of AIM envelopes is at Figure 4-25. The fire range is greater in a nose-on configuration than in a tail-on configuration. Indeed, the relative closing velocity (between the missile and the target) is greater in nose-on configuration than in tail-on configuration. The missile envelope is wider at high altitude than at low altitude because air density (and thus aerodynamic drag on the missile) decreases with the altitude. Two types of missile envelopes can be calculated by the FCC: the envelope for a nonmaneuvering target and the envelope with escape maneuver. When a target is engaged, a symbol is attached to it. This symbol represents all positions where the target may be after the missile’s flight. When the symbol attached to the target is totally enclosed in the missile’s envelope, the missile can be fired without the targets’s ability to normally avoid it. The pilot will usually prosecute using HUD symbology for intercept and launch info. Figure 4-23 shows a typical STT semi-active missile scenario.

146

CHAPTER

4

Fire-Control Radar

FIGURE 4-26 ¢ Typical HUD Air-toAir Cannon Symbology [Annotated Simulation Screen Capture. Lock-on Modern Air Combat http://lockon.co.uk/ en/modern_air_ combat/].

4.2.3.2 Cannon Mode Automated air-to-air gunnery fire control requires sensor inputs that can define the target in 3D space (azimuth, elevation, range) and calculate the target’s relative velocity vector with input of the ballistic characteristics of the round being fired. The solution results in the prediction of a point in space that the enemy aircraft will occupy after the predicted time of flight of the cannon round and providing the pilot a visible indicator (gun sight or HUD) by which to aim the cannon. Most fighter aircraft aim the cannon by maneuvering to superimpose the HUD generated pipper with the designated target (Figure 4-26). Gun accuracy is typically limited by recoil and barrel dispersion but not radar accuracy. 4.2.3.3 Air-to-Ground Missile Mode Unlike air-to-air missiles designed to hit highly maneuverable targets, air-to-ground missiles are designed for strikes on fixed targets or not very mobile targets. This is why the modern air-to-ground missiles use INS or GPS as their main guidance on geodetic coordinates. This guidance principle enables precision strikes at relatively low cost. Indeed, no seeker is needed for strikes against fixed targets. These weapons fired from standoff distance by day or night and in all-weather conditions offer a range exceeding tens of kilometers. 4.2.3.4 Air-to-Sea Missile Mode Like ground targets, sea targets are not very maneuverable. These targets are large and robust; they require a powerful warhead to be destroyed and thus a large missile. The sea targets are often detected at very long range, so the weapon shall also be long range. A typical air-to-sea missile operates as follows.

4.2

Airborne Fire-Control Radar

After its launch, the missile stabilizes in the direction of its target at its first cruising altitude, which is low enough to avoid detection by its target yet high enough to allow its active radar seeker head to acquire the target. This acquisition may be assisted by a target’s course corrections coming from the FCR via a suitable link. Midcourse guidance is by an INS or GPS navigation system and a radio altimeter, allowing the missile to fly a sea-skimming trajectory to its target. This sea-skimming trajectory prevents radars on board the ship to detect the missile too early. Then it descends to its second cruise altitude for the terminal phase, with a final approach at an altitude determined by prevailing sea conditions, sometimes as low as 3 m. Terminal guidance is provided by an active radar seeker. An ISAR image of the target (from the FCR) may be transferred into the missile computer. During the terminal phase, the missile seeker performs target imaging, comparison with the prestored ISAR image, and aim-point selection.

4.2.4 E-SCAN FCR Two kinds of electronic scanning are seen worldwide: 1. The passive electronically scanned antenna (PESA) radar and 2. AESA radar. The common feature of all E-SCAN radars, whether they are passive or active, is there are no motors or gimbals to point an antenna dish to search for or point at targets. The transmit-and-receive signal is steered electronically. Only the method for beam forming and the method for controlling the electronic scanning are different between PESA and AESA radars. E-SCAN allows very rapid changes in the signal direction and shape, making it possible to do things nearly impossible with physically pointed antennas.

PESA Radars PESA radars were developed before AESA radars. Indeed, the technologies required for AESA have only been affordable over the last decade or so. The first PESA FCR was Zaslon aboard the MIG-31 aircraft. This radar entered service in the early 1980s in the former Soviet Union. The radiation of the antenna takes place through about a thousand ferrite phase shifters that control the direction of the wave front. Although the concept and the technology of radar (mainly analog signal processing) have become outdated, this PESA enabled a multitarget radar without limiting the angular spacing of the targets traditionally encountered even with more modern M-SCAN radars. This E-SCAN based on ferrite phase shifters continues to be used on newly designed modern Russian FCRs. The first series of RBE2 radars installed on the French Rafale fighter are also fitted with PESA. E-SCAN is not performed by using ferrite phase shifters but is carried out by two electronically controlled deflection lenses located in front of a fixed illuminator (Radant technology). One lens controls azimuth steering, and the other controls elevation steering. PESA has also been used on other kinds of radars than FCR. AESA Radars AESA technology has been an important development for the radar industry. AESA radars are being procured in increasingly large numbers, for ground-based, sea-based, and airborne uses. In some early cases, operational radars

147

148

CHAPTER

4

Fire-Control Radar

FIGURE 4-27 ¢ Examples of PESA – Left: Russian ZASLON [http://www.airforcetechnology.com/projects/mig-31/images/2-mig-31-foxhound.jpg þ own work (Kemkemian)], Right: RBE2-PESA [(Courtesy THALES) Media resources Thales Airborne Systems].

were upgraded with new AESA antennas. But newer radars are being designed from scratch with AESA technology as a key part of the architecture. AESA rada rs have fixed, flat antennas made up of a large number of transmit-andreceive modules (Figure 4-27). The transmit-and-receive signal is steered electronically by controlling the phase of the signal at each of hundreds – even thousands – of modules to cause the wave front to add or subtract in space in such a way as to shape and point the beam. Compared to systems using M-SCAN or passive E-SCAN (PESA) where the transmit signal is generated by a separate source and applied to the antenna, the active array generates the transmit power and receive capability in each individual module. This distributed arrangement significantly reduces RF losses compared to systems with central transmitter and receiver. At a given overall power consumption, a figure of merit of the RF part (and thus performance in terms of detection range and tracking) is found to be much better than achieved by centralized systems. Because there are no moving parts to wear out and the antenna remains operational even if some modules have failed (graceful degradation), the antenna lifespan can run into the thousands of hours. Another advantage of such a distributed transmit-and-receive scheme is the ability to reconfigure the antenna’s subarrays according to the modes. This capability will be leveraged with the upcoming generation of AESA radars with multiple subarrays. AESA technology is most prevalent with airborne radars. While the operating techniques described for the various modes remain essentially the same, they can often be accomplished more quickly and the information obtained can be more accurate and more useful. Multiple modes can operate at the same time, something not possible with conventional pointed systems. The cost of producing the antennas and price of the powerful processors needed is such that AESA is not always cost-effective for many ground sensors. Airborne systems, however, are another story. Designers are finding the price worth the increase in capability; however, the reliability of the radar being significantly improved by the AESA, the cost of ownership is reduced. Table 4-1 summarizes the main features of AESA radars compared to older ones. All major air forces are adopting AESA radars for their next-generation aircraft. In addition to vastly improved radar performance, the radars can be adapted to be an important part of an overall situational awareness and electronic combat electronics suite.

4.2 TABLE 4-1

¢

Airborne Fire-Control Radar

Comparison of M-SCAN, PESA, and AESA

Motors, gimbals Transmission, reception Reliability RF losses Instant beam agility Simultaneous modes ability Multiple beam shapes ability Reconfigurable array

M-SCAN

Passive E-SCAN

Active E-SCAN

Yes Centralized Reference Reference No No No No

No Centralized þ – Yes Yesa Yesa No

No Distributed þþþ þþþ Yes Yes Yes Yes

a

Depending on available computational power.

FIGURE 4-28 ¢ Examples of AESA – Left AN/APG77 [(Courtesy Northrop-Grumman) http:// 132.228.182.15/solutions/f22aesaradar/assets/apg77.jpg], Right RBE2-AESA [(Courtesy THALES) Media resources Thales Airborne Systems].

The United States is not the only country working on airborne AESA radars. At the time of this writing, Russia and other European countries already have operational systems or planned systems for their new, frontline fighters. Figure 4-28 displays two examples of AESA.

4.2.4.1 Technological Aspects A typical transmit-and-receive module’s architecture is displayed in Figure 4-29. A transmit-and-receive module (TRM) contains mainly three parts: 1. a transmission stage, 2. a low-noise reception stage, and 3. a core chip that is controlled by either an FPGA or an ASIC. At transmission, the low-level incoming signal passes successively through: ●

an attenuator and phase shifter,

a driver amplifier (DRA),

a high-power amplifier (HPA), and

a circulator before finally reaching the radiating element.

149

150

CHAPTER

FIGURE 4-29 ¢ Synopsis of a Transmit-andReceive Module [(Courtesy THALES) Thales Airborne Systems (Lacomme)].

4

Fire-Control Radar

CORE CHIP

DRA

θ

HPA

Control FPGA or ASIC

Antenna

ATT LNA

MFC (Phase Shifter Gain Control)

FIGURE 4-30 ¢ Example of FirstGeneration TRM (late 90s) [(Courtesy THALES) Thales Airborne Systems (Lacomme)].

ASIC

Limiter

Limitor-LNA

Driver

Circulator

HPAs

At the reception, the signal from the radiating element is applied to the Low Noise Amplifier (LNA). A limiter is placed before the LNA to protect it against strong signals and to prevent its destruction. After the LNA, the signal passes through the attenuator and the phase shifter. The TRM currently in use relies on GaAs technology (Figure 4-30). The current trend is to ultimately replace GaAs by the following [3]: ●

GaN technology for transmitting and receiving will allow better efficiency of the active antenna by allowing: &

&

&

more peak power so that waveforms are easier to implement for a given RF average power or more average RF power if a supply power is available; a higher supply voltage (typically 32V), which simplifies the DC wiring of the antenna and the power supplies (lower DC intensity) as well as the impedance matching to attack the radiating element; and improved robustness of the LNA against strong signals so that a protective device will be easier to realize.

SiGe technology for the core chip that will allow miniaturization and lower cost.

4.2

Airborne Fire-Control Radar

151

4.2.4.2 New Modes Enabled by E-SCAN and AESA Two kinds of new modes and functions are present on the latest generation of AESA radars: 1. Modes and functionalities that can be implemented only because E-SCAN is present. 2. Modes and functions that do not directly depend on the presence of E-SCAN but whose feasibility is now made possible at affordable cost by the availability of some new technologies. These technologies, mainly driven by the explosion of civilian digital electronic market (mobile telecoms, the Internet, powerful computing servers, personal computing, etc.), are now available ‘‘off the shelf’’ and offer: ●

high-power real-time computing;

digital signal synthesis with high spectral purity;

Digital receivers, high speed Analog to Digital Converters (ADC), and so on; and

efficient and powerful simulation tools that dramatically reduce the duration and the cost of integration, verification, validation, and qualification (IVVQ) process of complex modes, complex system architectures, and so on.

In the United States and many European countries, the active antenna for FCR is now fully mature and can take advantage of these technologies from the civilian market to operate radar with more performing modes than were possible a few years ago. Typically, the new modes and functionalities available on the current generation of AESA are described hereafter.

4.2.4.3 E-SCAN-Only Functionalities E-SCAN-only functionalities are mainly: ● ●

the search while scan air-to-air mode and the ability to carry out simultaneous modes – for instance, air-to-air mode while an air-to-surface task is being performed.

Search While Track (SWT) As previously discussed, to carry out an air interception, the M-SCAN radars have several air-to-air modes, which are successively used (VS, RWS, and TWS, and then STT). The SWT mode, which requires a fast beam-steering agility (thus E-SCAN), merges all these previous modes into a single mode (SWT). A typical SWT mode is represented in Figure 4-31.

Data-Link

Weapon Quality Track (Wqt)

VS/RWS/TWS (Search Domain) WQT Track (TK)

FIGURE 4-31 ¢ Search While Track Mode for E-SCAN Radar [(Courtesy THALES) Thales Airborne Systems þ own work (Kemkemian)].

152

CHAPTER

Fire-Control Radar

4

P

1.0 PT

0.8 PD

PT

0.6 0.4

PC Electronic Scanning

Mechanical Scanning

0.2 D 0.2

0.4

0.6

0.8

1.0

FIGURE 4-32 ¢ Two Steps Detection Gain [(from [2]) Manuscript of ‘‘Air and Space borne Radar Systems – Ph. Lacomme, J.Ph Hardange, J.C. Marchais, E. Normant. SCITECH Publishing’’ (Lacomme with permission)]. ●

Principle of SWT: If no target is yet tracked, SWT operation performs the following: 1. The radar periodically sweeps a given search domain (in blue) like an M-SCAN radar. The waveform may be a VS mode if the focus is on the detection range (but without range information at this step) or of an RWS mode if range information is desired. 2. Once a target is detected at least once, there are two options: (a) The system decides the target has a high priority and triggers a fast opening sequence of a new track (TK) or new weapon-quality track (WQT). Thanks to the E-SCAN, a dedicated beam steering is associated with this track that can be afterward tracked regardless of its angular position provided it is within the whole AESA coverage. (b) The system decides this target has a low priority and it opens a new TWS track just as an M-SCAN would do. In this case, the target should remain within the search domain to be tracked. No particular antenna pointing is associated with this track, and it is refreshed each time the antenna’s scanning passes over it. In the case of the first option (high-priority target, TK, or WQT), the tracking initiation sequence consists of N successive repointings. The track is confirmed if it is detected at least K times out of N repointings. Figure 4-32 shows this detection strategy in two steps, providing, through the E-SCAN, a tracking range with a given probability PT(D) almost equal to the detection range with a cumulative probabilityPC(D) ¼ PT(D).5 Returning to Figure 4-32, we see that a given M-SCAN radar reaches PC ¼ 0:5 at the normalized distance 0.75. At this distance, the single-scan probability of detection is only 0.1, which is not sufficient to initiate a track with a conventional TWS scheme. A probability of tracking equal to 0.5 is reached at the normalized distance of 0.55. Regarding the ‘‘double-step’’ strategy of track opening permitted by the E-SCAN, PC ð0:75Þ ¼ 0:5 and PT ð0:75Þ ¼ 0:4. In this example, this means that the conditional probability of confirming a track at D ¼ 0.75, if a detection was obtained in the search phase, is 0.4/0.5 ¼ 0.8. In the case of E-SCAN, resulting in PT ¼ 0:5 at D ¼ 0.72.

The cumulative probability PC ðDÞ is the probability for a closing target to have been detected at least one time at ranges greater than D.

5

4.2

Airborne Fire-Control Radar

This example highlights a very significant improvement provided by the double-stage strategy of track initialization. Comparing the TWS scheme and the SWT scheme, the tracking range’s ratio is 0.72/0.5 ¼ 1.44. This tracking range’s improvement is equivalent to 6 dB. The increase of the figure of merit of the AESA radar’s RF front end compared to older M-SCAN or PESA radars and the processing gain provided by this detection strategy are the keys to the dramatic increase in operation ranges provided by these new AESA radars. ●

Target tracking strategy in SWT: When several targets are being tracked, an efficient management of time’s resources needs to be set up for the targets tracked with dedicated repointing. Indeed, without an efficient allocation of the time for the tracking and the time for the search phase, this tracking’s scheme loses its efficiency. Indeed, the time spent to update the tracks (TK or WQT) slows down the period of the search phase, thus reduces its efficiency. However, the system is able to track as many TWS targets as desired without these numerous tracks impact overall performance. The only limit to this is the computing power, but it is no longer a major problem with modern radars. Regarding the tracks that use dedicated repointing (TK and WQT), there are two main parameters to manage: &

&

The dwell time on each target direction during a repointing. The dwell time is mainly driven by the required signal-to-noise ratio. A distant target will require a long dwell time, but a close one will require only a short dwell time. The update rate. This parameter impacts mainly the track quality. Distant targets, far from the weapon range envelope (TK), do not need fast update rates. Conversely, a target close or within the firing envelope (WQT) requires accurate tracking even in the presence of evasive maneuvers on the part of the target. At constant track quality, the new trajectory estimation algorithms (IMM, etc.) enable relaxing the constraints on the update rate.

Simultaneous modes and functions An impossible thing to do with M-SCAN radar is to have, viewed from the operator, two radars in one: one air-combat radar and one Surface Combat radar. With the E-SCAN and now available processing capabilities, this old dream is now possible with a single radar and so a single platform. Tracking

Search

Simultaneous Modes

Air-to-Ground (Terrain Following)

FIGURE 4-33 ¢ Simultaneous Modes and Functions Enabled by E-SCAN [(from [2]) Manuscript of ‘‘Air and Space borne Radar Systems – Ph. Lacomme, J.Ph Hardange, J.C. Marchais, E. Normant. SCITECH Publishing’’ (Lacomme with permission)].

153

154

CHAPTER

Level 0

FIGURE 4-34 ¢ From Set of Sensors to Multifunction Sensors [Own work (Kemkemian, Nouvel, Chamouard) also published in IEEE papers].

FCR

Fire-Control Radar

4

Level 1

Level 2

Level 3

FCR

FCR

FCR

Fusion EWS

EWS

FCR & EWS

Fusion EWS

Level 4 FCR & EWS

EWS

FCR & EWS FCR & EWS

Time

A time sharing between tasks is established and the antenna beam is instantaneously switched from a direction for a task to another direction for another task (Figure 4-33). As trade-offs must be made, according to the mission objectives, the ‘‘system’’ manages the scheduling (for instance air-to-air and air-to-ground task with priority to the air-toground or conversely).

4.2.4.4 Other Modern Functionalities Not Directly Related with E-SCAN System Integration The next generation of AESA long-range radar and integrated sensors suite are designed from the beginning as a system of sensors and not as a set of stand-alone sensors. The successive steps are summarized in Figure 4-34. This figure (ref. [1]) illustrates the evolution of a set of sensors (radar & EWS) toward a multifunction system of sensors. It can obviously be extended to the electro-optical sensors (radar & EWS & EOTS), even to the communication means of each platform. Radars discussed in this section correspond to ‘‘level 2’’: ● ●

data fusion at system level and close exchange of data/signals between sensors via ad hoc links for mutual enhancement.

The targeting sensors are designed to assure that the pilot concentrates on the combat environment, not on the onboard systems. Functions are heavily automated. Designers concentrated on a high level of systems integration and data fusion with targeting data developed from the AESA radar, E/O targeting system, Electronic Warfare suite, and the communication, navigation, and identification suite. This long-range sensor provides all-weather, standoff target detection, minimizing threat exposure. The radar will probably be the first onboard sensor to identify a target, but the entire suite will be used to prosecute whatever mission is called for.

New Capacities in Air-to-Air For air-to-air operations, up to date radars, like APG-81 and other modern AESA radars support such features as passive search and multitarget, and beyond-visual-range tracking and targeting. It also will support a cued search feature, in which the radar is cued toward another sensor’s line of sight. That other sensor can be on-board, off-board or pilot-directed. An example of cued radar’s search by the ESM part of the EWS is shown in Figure 4-35. New Capacities in Air-to-Ground These new AESA radars come with a SAR terrain mapping function for air-to-surface surveillance and targeting. It is said to be comparable to the terrain mapping radar used in reconnaissance aircraft, unmanned air vehicles, and the E-8C Joint Surveillance Target Attack Radar System (Joint STARS) aircraft.

4.2

RESM D

Airborne Fire-Control Radar

OA #1

RESM DOA #2

155

Small Scanning Area Non Transmitting FCR

FIGURE 4-35 ¢ Cued Search in Air-to-Air [Own work (Kemkemian, Nouvel, Chamouard) also published in IEEE papers].

FIGURE 4-36 ¢ SAR Image Produced by Modern AESA Radar [(Courtesy THALES) Thales Airborne Systems þ own work (Kemkemian)].

Designers worked to provide these new radars with higher resolution for easily recognizable features on the ground and to have the radar look at three or four times more terrain than previous radars (Figure 4-36).

New Capacities at the Level of Human Machine Interface (HMI) and System Automation By activating buttons around the aircraft’s panoramic MultiFunction Display (MFD) the pilot can select from the radar’s many software-driven modes: target identification and tracking, air-to-air, air-to-ground, air-to-sea surface target detection and electronic warfare, as well as SAR ground mapping. It can designate both ground targets and airborne targets simultaneously (thanks to the ability to carry out both air-toair and air-to-ground tasks enabled by the AESA). Describing a possible mission, as the tactical aircraft enters the combat zone, an indication of a potential target pops up on the pilot’s MFD. Flying toward the area of interest, the pilot presses his or her finger against the touch-screen display and views a much clearer, close-up image of the target, which is identified as an enemy ground vehicle. The pilot presses the screen again, and target designation and weapons status imagery appears on the visor of his helmet-mounted display (HMD). Closer in, he views the target, now being automatically tracked; the crosshairs in the visor lock on to the

156

CHAPTER

4

Fire-Control Radar

target; and the pilot fires a missile that follows a laser beam to its destination. The pilot again views the MFD and notes that the target has been destroyed. The idea is to simplify the pilot’s mission, making for easy target detection and destruction of ground vehicles, ships, or enemy aircraft. The complexity lies in the sensors, displays, and massive processing power that make up the automatic targeting capabilities of a 21st-century, multibranch, multinational, multimission supersonic fighter. The fused targeting data can be overlaid onto a battlefield situation display that the pilot has uplinked from a ground base or another aircraft. The intent is to produce battle scene awareness to support an Observe, Orient, Decide and Act (OODA) sequence for pilots.

4.2.5 The Future of FCR Sensor Improvement The future AESA will be fitted with more than the four traditional quadrants. These multiple subarrays will allow ABF for efficient multiple jammers cancellation, space–time adaptive processing for clutter cancellation and detection improvements. In Europe, such AESA were pioneered by the tri-national AMSAR program (Airborne Multirole Solid-state Active array radar) [4] involving SELEX (formerly BAe Systems) for the United Kingdom, CASSIDIAN (formerly EADS Defense) for Germany, and THALES for France. The latest AMSAR’s AESA configuration was made up of eight subarrays (Figure 4-37). By using multiple subarrays, the processing enables better protection against attacks, better detection of slow moving targets in GMTI while keeping accurate angle localization. The number of subarrays determines the degrees of freedom’s number for interference cancellation. In the worst case, to clean the Sum channel and the two Difference channels (for monopulse measurements), it is required that: NSA NJ , where NSA and NJ are respectively the number of subarrays and the number of jammers. ●

FIGURE 4-37 ¢ AMSAR Subarrays [Thales Airborne Systems þ own work (Kemkemian)].

Adaptive beam forming for multijammers cancellation – This processing is described in section 4.4.2. STAP for clutter cancellation – STAP consists in forming a two-dimensional filter (Angle-Doppler) so that for each Doppler frequency (corresponding to a possible moving target), the antenna pattern has a null in the direction of fixed clutter having the same Doppler. An example is given in Figure 4-38. The clutter locus in 2D is represented by the oblique black ridge. The targets are in grey. Some of them (the slowest ones) can only be visible in a 2D representation since they are hidden by Guard Channel Assembly

4.2

Airborne Fire-Control Radar

Optimal Angular Filter sin (q )

Optimal Time Filter Fd

157 FIGURE 4-38 ¢ STAP Filter (Side Looking Radar) [Thales Airborne Systems þ own work (Lacomme)].

Fd = 2V sin (q ) λ

Range

Range

Optimal STAP Filter

Targets

Doppler Range / Doppler Map Conventional Processing

Doppler Space-time Adaptive Processing

the clutter in 1D optimal filter (either in time/frequency domain or in space/angle domain): &

&

&

&

To detect only these targets, two degrees of freedom are needed (two subarrays) if there is no range/Doppler ambiguity. To localize these targets in azimuth (GMTI case), three degrees of freedom are needed (always without range/Doppler ambiguity). To localize these targets both in azimuth and elevation (air-to-air case), it is nice to have six degrees of freedom. Waveforms with range or Doppler ambiguities require extra degrees of freedom.

More details on STAP can be found in Volume 2, Chapter 10. An example of STAP benefit in GMTI is given in Figure 4-39. The slow targets, not detectable with conventional Doppler processing, are visible after STAP. Another example of STAP application is given for MPRF mode in air-to-air (Figure 4-40). Due to the Range and Doppler folding, the whole Range-Doppler domain is desensitized by the clutter returns. With STAP processing the clutter þ noise level in the target’s domain (excluding very slow velocities with respect to the ground) is now about the thermal noise only level.

FIGURE 4-39 ¢ STAP Benefit in GMTI [(from [4]) AMSAR – A European Success Story in AESA Radar – J.L. Millin, S. Moore, W. Bu¨rger, P.Y. Triboulloy, M. Royden, J. Gerster. IEEE international Radar Conference, October 2009].

FIGURE 4-40

–1.00 –1.00 –0.75 –0.50 –0.25 0.00 0.25 0.50 0.75

–0.75

–0.50

–0.25

0.00

0.25

0.50

0.75

Antenna Pattern

¢

0 –150 –200 –400 –500 –600 –700

dB

–160

–155

–150

–145

–140

–135

–130

–125

–120

–115

–110

–105

Ambiguous Velocity

Clutter/Noise STAP Channel

STAP Benefit with MPRF Waveform [Thales Airborne Systems ].

Ambiguous Velocity

Clutter/Noise Sum Channel

dB

5

10

15

20

25

30

35

40

45

50

4

Ambiguous Range

CHAPTER

Ambiguous Range

158

Fire-Control Radar

4.2 ●

Airborne Fire-Control Radar

159

Multifunction sensor: The radar is now a multipurpose cooperative sensor. In conjunction with the EWS and the electro-optical targeting system, it is part of a system of cooperating sensors. For example, the radar AESA can be used to perform more efficient EW functions in its bandwidth. In another area, the powerful AESA of the radar, in conjunction with a modem, can carry out ‘‘ad-hoc’’ data-links at long range and very high speed.

Key to the futuristic fighter’s targeting capability is sensor integration and data fusion. Mission system software fuses data from: ●

the electronically scanned array radar,

EOTS with FLIR and IRST system,

the electronic warfare suite, and

the communication, navigation, and identification (CNI) suite, providing identification friend or foe (IFF) and off-board data delivered via a high-speed data link.

All these sensors are arranged as a set of distributed apertures, both EM and ElectroOptical (Figure 4-41). ●

Smart management of the system of sensors – The goal is for the pilot to receive unprecedented situational awareness from a mission systems package that incorporates modular open systems architecture, object-oriented design and common commercial off-the-shelf (COTS) processors. Combined with onboard precision weaponry–missiles, smart bombs, and a cannon, the future fighter will ‘‘compress the kill chain,’’ and have the ability to destroy targets ‘‘within single-digit minutes’’ of their detection.

In a typical scenario, the pilot would first detect a beyond-eyesight target in a predominantly radar image on the MFD. As the target gets closer, the EOTS imagery automatically creates a clearer picture of the target on the MFD. At this point, the pilot assesses an operational picture of the battle space, evaluates the threat responses, and rapidly plans a route to secure minimum exposure and maximum weapon effectiveness, as well as determining the best choice of weapon. Once the decision is made to attack the target, the pilot would switch from the headdown to the head-up display in his or her helmet-mounted visor. In addition to presenting a center cross that locks onto the target for a point-and-shoot capability, the FIGURE 4-41 ¢ Example of a Conformal Broadband Multifunction Array (radar, ESM, COM) [Thales Airborne Systems].

160

CHAPTER

4

Fire-Control Radar

HMD also presents the status of available weapons, a symbol for IFF and indication of the target’s range, closure, and velocity. With most of the target detection and presentation achieved automatically, the OODA process, from acquisition to destruction, can be done within a few minutes. The targeting sensors and processors will make these fighters not just a combat aircraft firing weapons, but a first-day-of-the-war, multimission aircraft able to perform autonomously, cooperatively or remotely, using information from off-board sources. In a cooperative mission, the integrated sensor suite would package and format targeting data to form a waveform for delivery by the CNI to a ground base or other aircraft via data link. ●

Network of cooperative systems of sensors – The combat platforms present in the theater of operation, fitted with multifunctional sensor systems are now networked. A global tactical situation is established through the contribution of each individual platform, and then is shared by all platforms thanks to interoperable data-links. The theater of operation contains many platforms. Some of them operate in active mode (radar), the others may be operated in passive mode (Multistatic receive only radars sensors, ESM sensors, Electro-Optical sensors, etc.). Such a network has many advantages: &

&

&

&

&

Difficult to neutralize: the loss of a node does not mean the loss of the whole network (same concept as Internet). Diversity of observation: a target is observed from many angles of aspect. The traditional weaknesses of radar no longer exist, such as the Doppler notches. Reduction of the exposure to threats of manned and costly platforms: UCAV and UAV may be sent to hazardous areas without endangering crews. Difficulty for the enemy to identify the origin of the imminent threat for him. Spectral resources saving: the minimal number of platforms is in active mode. The other ones remain in silent mode. An example of networked system of cooperative platforms is shown in Figure 4-42. It is obvious that the key elements for deploying such a network of sensor are:

high-speed, long range communication systems; and

very accurate synchronization means.

4.3

SURFACE-BASED FIRE-CONTROL RADAR

Technological developments of surface-based FCR are similar to those of airborne FCR. This particularly concerns electronics and signal processing. Depending on the implantation of these radars, space constraints are more or less critical.

4.3.1 Surface-to-Air Fire-Control Radar 4.3.1.1 Antiaircraft Artillery (AAA) AAA can be employed as either unguided ‘‘barrage fire’’ or sensor guided ‘‘directed fire.’’ One of the earliest uses of radar as an AAA aid was by the German air defense in World War II when the altitude of incoming Allied bomber formations was measured to allow accurate setting of Flak projectile fuses.

RELAY or C3I

HDL

MDL

Fighter 4

3

FIGURE 4-42

C3I

¢

FEBA

UCAV REACQUISITION 4 TARGETING FIRING

SAM

UCAV

HIGH RATE DATA LINK (HDL)

4

1

SAM

MALE RECONNAISSANCE

SURVEILLANCE

UCAV

HALE

Cooperative Network of Platforms [Thales Airborne Systems (Lacomme)].

STRIKE DECISION

MEDIUM RATE DATA LINK MDL

RECONNAISSANCE IDENTIFICATION

2

SAM

4.3

Surface-Based Fire-control Radar 161

162 FIGURE 4-43 MK-15 CIWS.

CHAPTER

4

Fire-Control Radar

¢

Search Radar

Track Radar

More modern AAA systems employ radar sensors allowing accurate track predictions. The weapon system will fire at a point in space that the target is predicted to occupy after a predetermined time of flight of the projectile. As an example, the MK 15 Phalanx (RAYTHEON) ‘‘Close-In Weapons System’’ or CIWS (Figure 4-43) is a fastreaction, rapid-fire 20-millimeter gun system that provides US Navy ships with a terminal defense against antiship cruise missiles (ASCM) that have penetrated other fleet defenses. This system is designed to engage ASCM and fixed-wing aircraft at short range. The Phalanx automatically enables ‘‘kill chain’’ functions usually performed by separate, independent systems such as search, detection, threat evaluation, acquisition, track, firing, target destruction, kill assessment, and cease fire. The fire-control assembly is composed of a search radar (Ku-band, digital Moving Target Indicator: MTI) for surveillance and detection of hostile targets and a track radar (Ku-band, pulse Doppler monopulse) for aiming the weapon while tracking a target. The unique closed-loop fire-control system that tracks both the incoming target and the stream of outgoing projectiles gives CIWS the capability to correct its aim to hit very fast-moving targets. The monopulse beam design enables the azimuth and elevation tracking accuracy required in the AAA environment [5]. Another defense system against antiship cruise missiles is the ‘‘Goalkeeper’’ (Figure 4-44, THALES Nederland). As the Phalanx system, the Goalkeeper is constituted of a search radar and a tracking radar. The X-band search radar can handle up to 18 targets at once. The tracking radar operates both in X and Ka bands. This dualband operation ensures: ● ●

good robustness against CME, and double beamwidth (the Ka band beam performs an accurate angle tracking while the X-band beam allows a fast acquisition [from the search radar data] or a fast reacquisition).

4.3

Surface-Based Fire-control Radar

163 FIGURE 4-44 ¢ Goalkeeper System [http://fr.wikipedia. org/wiki/ Goalkeeper_CIWS].

4.3.1.2 Surface-to-Air Missile Systems These systems, designed to engage and destroy airborne threats, are generally made of two subsystems: 1. A first subsystem fitted with a search radar having a large coverage in elevation and a coverage of 360 in azimuth. Comparing the requirements of the search function of an airborne FCR and the requirement of the search function of a surfacebased FCR, the following differences are noted: ●

The required coverage in azimuth is 360 , while an airborne FCR covers the front area. This requirement is not really an issue with mechanically scanned systems. However, the evolution toward AESA systems requires either an extra mechanical rotation axis in azimuth or a set of facetted AESA. In the future, systems with cylindrical AESA (i.e., non planar arrays) are being studied. The angular domain in elevation is often quite large. Generally, the requirement is to provide oversight of a ‘‘quasi’’ cylindrical area whose cut is shown in Figure 4-45

It is interesting to note, if q is the elevation angle, the required detection range is: RD ¼ AB cscq < RA . That is why the first air-surveillance systems were using an antenna whose radiation gain pattern (called ‘‘cosecant’’) varied with elevation in order to adjust the sensitivity according to the required detection range. It is a simple solution, but it does not allow very accurate localization in elevation. Also, new modern E-SCAN radars are adopting another strategy: ● ●

A quick elevation E-SCAN is achieved during the antenna’s rotation in azimuth. The antenna gain does not vary during the quick scan; however, the dwell time on target decreases with elevation according to a cosecant law. Thanks to E-SCAN, this method provides a fast refresh time while providing a good angular accuracy.

2. A second subsystem equipped with a tracking radar with more accurate azimuth and equation accuracy. The tracking radar is cued by the search radar when it detects a target, then it tracks the target while providing the missile guidance. The second subsystem contains also the missiles battery. Again, tracking systems with M-SCAN antenna

164

CHAPTER

4

Fire-Control Radar C

FIGURE 4-45 ¢ Typical Search Domain of Surfaceto-Air Systems [Own work (Kemkemian)].

B max. Flight Level

Radar

A H Some Tens of km

FIGURE 4-46 ¢ HAWK System. Left: PAR, Right Top: HIPIR, Right Bottom: Missile Battery [(from Wikipedia) http://en. wikipedia.org/wiki/ MIM23_Hawk þ own work (Kemkemian)].

only allow to engage one target at a time. A tracking system with E-SCAN radar, combined with active homing missiles (active radar seekers), allows multiple targets engagement. These subsystems can be made up of separate units or be grouped on a single platform. Two representative systems are the HAWK system and the CROTALE system:

Improved HAWK (I-HAWK) System A Hawk unit uses several different ground radars and control systems. A typical I-Hawk battery mainly consists of (Figure 4-46): ●

1 I-PAR: Improved Pulse Acquisition Radar (AN/MPQ-64) – A search radar with a 20 rpm rotation, for high-, medium-, or low-altitude target detection. 2 HIPIR: High Power Illuminator Doppler Radar (AN/MPQ-61) – Which tracks designated targets and provides target illumination for the missile’s semi-active seeker.

Note: The old HAWK versions have an extra ‘‘Range Only Radar’’ (ROR) which is a K-band pulse radar intended to provide ranging data when the other radars are jammed by countermeasures. I-PAR is a Doppler (MTI) radar that helps separate targets from ground clutter. It operates in the C-band frequency range with a peak operating power of 1,000 watts. It is

4.3

Surface-Based Fire-control Radar

165 FIGURE 4-47 ¢ Crotale NG [http:// www.spyworldactu.com/spip.php? article8685 (original image: French Mod but no longer available)].

fitted with a rotating antenna in azimuth but electronically steered in elevation. The detection range is about 100 km versus 3 m2 target. HIPIR is an X band CW system which is used to illuminate targets in the Hawk Missile Battery. The unit comes mounted on its own mobile trailer. Unit automatically acquires and tracks designated targets in azimuth elevation and range rate. The CW operation requires simultaneous transmission and reception, so two separate antennas provide sufficient isolation.

Crotale NG System The whole set (search radar, tracking radar, and missiles battery) is grouped and mounted on a single trail (Figure 4-47). This system is composed of: ●

A pulse-Doppler search radar operating in S-band able to operate in motion.

A pulse-Doppler tracking radar operation in Ku-band with a narrow beam.

An Infrared Camera.

A visible Camera.

An infrared localizer intended for missile tracking.

4.3.2 Surface-to-Surface Fire-Control Radar These surface-to-surface systems are quite similar in their principles to the surface-to-air systems. This is essentially the nature of the targets which differentiates them.

4.3.2.1 Maritime Surface-to-Surface Fire-Control Systems Radar is a primary sensor in maritime surface-to-surface engagements. Sensors are only required to provide a 2D input (range and azimuth) to the fire-control system to produce a viable solution as targets are mainly subsonic or supersonic sea-skimming missiles. The elevation can usually be assumed to be 0 feet Medium Sea Level (MSL). Radar inputs can be used for providing fire-control solutions for surface-to-surface gunnery or to provide targeting information to antiship cruise missiles which have onboard radars for active homing guidance.

166

CHAPTER

4

Fire-Control Radar

FIGURE 4-48 ¢ AN/TPQ-47 FireFinder Radar [www. fas.org].

4.3.2.2 Fire-Finder Radar Fire-Finder radar systems detect and backplot the fire of adversary weapons, artillery, rockets, and mortars to provide pinpoint targeting information for the counterattack. The systems also correct and improve the delivery of friendly fire. Fire-Finder radars are tri-dimensional radars with 2D E-SCAN (Figure 4-48). The stationary antenna sweeps a rapid sequence of beams along the horizon, forming an electronic radar curtain over a 90 area. Any target penetrating the curtain triggers an immediate verification beam. Upon verification, an automatic tracking sequence begins. While tracking any single target, the radar continues scanning, locating, and tracking others. Signal and data processors test each track to filter out birds, aircraft, and other unwanted returns, giving Fire-Finder radars an extremely low false-location rate and a very high probability of location rating. Once the computer establishes a target’s validity, it ‘‘smooths’’ the measured track data, deriving a trajectory that it extrapolates to establish the target’s firing position and impact location. Those data are displayed on a visual map and printed out in map coordinates.

4.3.3 Principles of Missile Guidance Only the principles that are used in modern systems are discussed here.

4.3.3.1 Surface-to-Air Missile Engagement Envelopes A surface-to-air missile (SAM) is neither guided after a target nor pursues or chases an aircraft. Instead, the fire-control computer predicts an intercept point on some future part of the target flight path based on the known flight parameters from the target tracking radar and the known maneuverable envelopes of both the target and the missile (Figure 4-49). The missile is accelerated for the brief initial phase of its flight (the boost phase) after which it can never again speed up: It is accelerated toward the predicted intercept point, after which it is only capable of slight course corrections to keep it pinpointed on the point of impact.

4.3

Surface-Based Fire-control Radar

H, km 3

4 c

d

5

b

2

1

a

5 5

e 10

D, km

The effective engagement envelope of a nominal SAM system is determined by the location and speed of the airborne target. Maximum engagement ranges vary from a highspeed nonmaneuvering closing target to a high-speed crossing target, generating the highest miss distance due to required missile maneuvering. Minimum range is usually determined by the end of the missile’s boost phase when missile guidance is initiated. SAMs have specific engagement envelopes. Firing at targets within the heart of the envelope increases the likelihood of a hit. Just like air-to-air missiles, the envelope varies based on the target’s range, altitude, and aspect. In the preceding engagement diagram, the area defined by numbers 1 through 5 represents the missile’s effective area (assuming that the objective is motionless). This envelope shifts if the target is moving toward the launcher, in the area defined as a through e. In this case, the missile must be fired at longer range since the target will fly part of the way into the missile. If the missile is fired too late (once the target has crossed the a–b–c line), it passes out of the envelope before the missile arrives.

4.3.3.2 Missile Guidance Modes The different types of missile guidance used in SAM systems include: Command Guidance The target is tracked by an external radar (see Figure 4-50). However, a second radar tracks the missile itself. The tracking data from both radars are fed into a ground-based computer that calculates the paths of the two vehicles. This computer also determines what commands need to be sent to the missile-control surfaces to steer the missile on an intercept course with the target. These commands are transmitted to a receiver on the missile that allows the missile to adjust its course. An example of command guidance is the Russian SA-2 surface-to-air missile. Semi-Active Homing Guidance The target needs to be continuously tracked by the FCS, but no guidance data is directly sent from the FCS to the missile. The energy reflected by the target is intercepted by a receive-only radar (seeker) located in the missile’s nose (Figure 4-51). An onboard computer receives data from the seeker to determine the target’s relative trajectory and sends correcting commands to control surfaces so that the missile will intercept the target. This system is also sometimes referred to as bi-static, meaning that the radar waves that intercept the target and those reflected back to the missile are at different angles (and at different ranges) to one another.

167 FIGURE 4-49 ¢ Nominal SAM Engagement Envelope [(from [6]) http://www.scribd. com/doc/69703675/ 18/AIR-COMBATBASICS].

168 FIGURE 4-50 ¢ SA-2 Command Guidance System [Self-Developed Graphic (Ballard)].

CHAPTER

4

Fire-Control Radar Target

Target Tracking

Intercept

k

Uplin

Missile

Missile

Target Command Transmitter

Missile Tracking Missile Tracker

FIGURE 4-51 ¢ Sea Sparrow SemiActive Guidance [Self-Developed Graphic (Ballard)].

Computer

Target Tracker

RIM-7 NATO Sea Sparrow Semi-Active Guidance

Radar Waves from Launching Ship

Missile Reflected Radar Signals

Active Homing Guidance Active homing missiles typically use radar seekers to track their target. Once the seeker is locked onto the target, the missile becomes autonomous: the fire-control system does not need to continue illuminating the target or transmit guidance data to the missile. For this reason, active homing missiles are often called fire and forget. The advanced medium-range air-to-air active homing missile has been adopted for use in the surface-to-air role by mounting it on a high-mobility multipurpose wheeled vehicle (HMMWV) and integrating it into various air-defense networks for target cueing. The Norwegian advanced surface-to-air missile system and the U.S. Marine Corps’ complementary low-altitude weapon system (Figure 4-52) are two examples.

Surface-Based Fire-control Radar

4.3

169

FIGURE 4-52 ¢ HMMWV AMRAAM Launcher [SelfDeveloped Graphic (Ballard)].

HUMRAAM

Reflected Radar Signals

Radar Waves from Missile

Patriot Track-via-Missile (TVM) Radar Waves from Launching System

Target and Missile Information from Missile

link

Down

k

Uplin

Missile Missile Command Information to Missile

Reflected Radar Signals

Retransmission Homing Guidance A more unusual example of homing guidance is the retransmission method (Figure 4-53). This technique is very similar to command guidance but with a unique twist. The target is tracked via an external radar, but the reflected signal is intercepted by a receiver onboard the missile as in semi-active homing. However, the missile has no onboard computer to process these signals. The signals are instead transmitted back to the launch platform for processing. The subsequent

FIGURE 4-53 ¢ Patriot Track-viaMissile TVM Tracking [SelfDeveloped Graphic (Ballard)].

170

CHAPTER

4

Fire-Control Radar

commands are then retransmitted back to the missile so that it can deflect control surfaces to adjust its trajectory. This method is also sometimes called track via missile (TVM) since the missile acts as a conduit for tracking information from the target back to the ground control station. The advantage of TVM homing is that most of the expensive tracking and processing hardware is located on the ground, where it can be reused for future missile launches rather than be destroyed at missile detonation. A downside, though, is that the TVM method also requires robust high-speed communication data links between missile and control station, limiting the system to rather short ranges. Retransmission homing guidance is used on the Patriot surface-to-air missile system.

4.4 ELECTRONIC COUNTER COUNTERMEASURES The objective of ECCM techniques is to allow the accomplishment of the radar mission while countering the effects of the enemy’s ECM [7]. The main jamming techniques to be countered are described hereafter.8 These ECM – and thus also ECCM – apply to both airborne and surface-based FCR. Some ECM techniques are intended more against airborne FCR whereas others are aimed against surface-based FCR. So, some ECCM will be more prevalent on airborne systems and others will be encountered more on surface-based systems.

4.4.1 Noise Jamming This jamming aims at reducing the signal-to-noise ratio to prevent the target detection: ●

Barrage Noise – This jamming is carried out by a powerful jammer onboard a ‘‘stand-off’’ platform. It attacks the victim radar by its antenna’s sidelobes: the lower they are, the more difficult is the jamming. Escort Jamming – This jamming is performed by an escort A/C and attacks the victim radar by its antenna’s main lobe or by its first sidelobes. Self-Protection Jamming – The jamming is issued from the target itself and attacks the victim radar by its antenna’s main lobe. There is no possible angular discrimination between target and jamming.

4.4.2 ECCM against Noise Jamming The first ECCM consists in using an unpredictable frequency agility sequence to force the jammer to transmit over a broadband (the jamming power within the instant radar bandwidth is reduced). The jamming noise enters the victim radar through its sidelobes. It can be reduced by a Side Lobe Canceller (SLC). The SLC takes advantage of a ‘‘guard’’ antenna to form 0 a ‘‘cleaned’’ sum channel. The SLC computes: S ¼ S lA, where ‘‘S’’ and ‘‘A’’ are

8

Note: Only the main techniques that are efficient against modern radars have been discussed.

4.4

Electronic Counter Countermeasures

FIGURE 4-54 ¢ Adaptive Beam Forming for Multiple Jammers Cancellation [Thales Airborne Systems].

JAMMER

JAMMER

ADAPTIVE DETECTION RANGE CLEAR AIR DETECTION RANGE

CONVENTIONAL RADAR JAMMED DETECTION RANGE

171

JAMMER

respectively the sum channel and the guard channel. The coefficient l is adaptively determined to maximize the signal-to-noise ratio. The auxiliary antenna’s gain needs to be greater than the sidelobes gain otherwise the processing is not efficient. So far, SLC is mostly used on ground-based radars, but it will be extensively used to counter multiple jammers in future airborne radar with active arrays and multiple subarrays by using ABF (Figure 4-54). In the case of jamming entering through the edges of the mainlobe or through the fist sidelobes, the SLC is no longer helpful as the auxiliary antenna’s gain is not sufficient. On future AESA radars with multiple subarrays, the solution will be to use ABF (Figure 4-54). Adaptive beam forming is a generalization of the SLC: its principle is to adaptively form, by linear combinations of subarrays, ‘‘cleaned’’ channels for detection and angle measurements of the target. In the case of self-protection jamming, the previous ECCM are not very efficient, especially the cancellers using SLC or ABF which do not work at all. The target can no longer be detected and the distance of the target can no longer be measured (except at short-range where the Signal to Jamming Ratio becomes favorable to the radar, or by using a ‘‘Burn Through’’ look onto the target). However, the target remains localizable in angle by its jammer transmission (jammer strobe) and can be tracked in a degraded mode; moreover, a co-operative system with strobes from two or three spatially separated radars allows the jammer to be localized.

4.4.3 Gated Noise Jamming A noise area is generated around or near the target echo (in range and/or Doppler). This kind of jamming requires the synchronization of the jammer on the radar’s waveform.

172

CHAPTER

4

Fire-Control Radar

Such synchronization is quite difficult with non-Doppler radar waveforms with pulse to pulse frequency/PRI agilities.

4.4.4 Range and Velocity Deception This jamming consists of creating false targets around the true target to affect tracking: ●

Multiple false-target generation – Multiple false targets are generated to saturate the display and mislead tracking initialization on the ‘‘good’’ target or to overwhelm the processing. The ECCM prevents overloading of the processing. It should also be noted that all false plots have the same direction as the ‘‘true’’ target, so angle only tracking remains feasible. Range gate pull-off (RGPO) and velocity gate pull-off (VGPO) – First, a false strong echo is superimposed on the target’s skin echo. This false echo is then progressively moved away from the target either by distance (RGPO) or by Doppler (VGPO). The purpose of this jamming is to capture the tracking on the real target. When the track is moved far enough away from the target, the false echo is stopped, which leads to lost tracking.

4.4.5 Angle Deception These techniques aim to disrupt the angle measurements in order to lose the angular tracking: ●

Cross-eye – This jamming employs two spatially separated jamming sources. Each source acts as a repeater-type jammer transmitting the same signal at the same time but with a phase shift of about 180 to create a wave-front distortion at the victim radar’s antenna level; as a result, a strong ‘‘glint effect’’ occurs in the angle. Cross-polarization – First, the jammer measures the polarization of the victim radar’s antenna. Jamming then consists of transmitting a strong signal in cross-polarization; the S and D patterns are so altered that they seem to have been exchanged. However, the gain is reduced by 20 dB to 40 dB, depending on the antenna’s quality. As a jammer is unable to transmit a perfect cross-polarized signal, a high-quality antenna will not be vulnerable to such jamming.

4.5 THE ‘‘AN’’ EQUIPMENT-DESIGNATION SYSTEM It is useful to review the designation of U.S. military electronics and communications equipment that allows a quick recognition of fire-control radar when referenced by its designation number. The Joint Electronics Type Designation System (the AN system) consists of a three-letter designation followed by a number. The first letter designates the type of platform: ‘‘A’’ for piloted aircraft, ‘‘S’’ for surface ship, ‘‘T’’ for ground transportable, and ‘‘M’’ for ground mobile. The second letter designates the type of equipment: ‘‘P’’ is always for radar. The third letter designates the purpose of the equipment:

4.7

Further Reading

‘‘G’’ for fire control and ‘‘Q’’ for special or a combination of two or more functions. Several examples of the AN system are listed below: ●

AN/APG-77 – Pulse Doppler X-band multimode radar used on F-22 aircraft,

AN/SPG-62 I – J-band fire-control radar used on Aegis-class combatant ships,

AN/TPQ-37 – mobile ground artillery locating radar, and

AN/MPQ-53 – multifunction phased array radar used with the Patriot SAM.

4.6

REFERENCES

[1] Radar and Electronic Warfare Cooperation. How to Improve the System Efficiency? S. Kemkemian, M. Nouvel-Fiani, E. Chamouard. IEEE/AESS Magazine, August 2011, Vol. 26, number 8. [2] Air and Spaceborne Radar Systems. P. Lacomme, J.P. Hardange, J.C. Marchais, E. Normant. William Andrew; I edition (Dec 17, 2007) Norwich NY. [3] Thales Components and Technologies for T/R Modules. Y. Mancuso. IEEE International Radar Conference, October 2009. [4] AMSAR: A European Success Story in AESA Radar. J.L. Millin, S. Moore, W. Bu¨rger, P.Y. Triboulloy, M. Royden, J. Gerster. IEEE International Radar Conference, October 2009. [5] http://www.fas.org/man/dod-101/sys/ship/weaps/mk-15.htm [6] http://www.scribd.com/doc/69703675/18/AIR-COMBAT-BASICS [7] Introduction to Electronic Defense Systems, Second Edition. Dr. Filippo Neri. Artech House Boston MA.

4.7

FURTHER READING

Boyd: The Fighter Pilot Who Changed the Art of War. R. Coram. Little, Brown & Co. ISBN10:0316796883 Fire-Control Radar Fundamentals Naval Education and Training October 1, 2007 ISBN10:1466310766

173

CHAPTER

Airborne Pulse-Doppler Radar

5

Aram Partizian, Georgia Tech Research Institute

Chapter Outline 5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9 5.10 5.11 5.12 5.13 5.14

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Geometry. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Doppler Shift and Motivation for Doppler Processing . . . . Range and Doppler Distribution of Clutter . . . . . . . . . . . . . . . . Contours of Constant Doppler and Range . . . . . . . . . . . . . . . . Example Scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pulse-Doppler Conceptual Approach . . . . . . . . . . . . . . . . . . . . Ambiguities, Folded Clutter, and Blind Zones. . . . . . . . . . . . . . Overview of PRF Regimes . . . . . . . . . . . . . . . . . . . . . . . . . . . . High PRF Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Medium PRF Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Low PRF Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5.1

........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........

.... .... .... .... .... .... .... .... .... .... .... .... .... ....

175 177 181 185 196 199 203 216 226 228 235 246 248 249

INTRODUCTION

Airborne pulse-Doppler radar is used to provide detection and tracking of airborne moving targets in the presence of clutter. The pulse-Doppler waveform consists of a coherent burst of pulses, generally with constant pulse width and pulse repetition frequency (PRF). The returns from the environment are received, range-gated, and coherently integrated to form a Doppler spectrum for each range gate. The receiver– signal processor divides the Doppler spectrum into narrowband Doppler filters to isolate the moving target from competing clutter on the basis of their differences in Doppler shift and, therefore, radial velocity. Two other radar modes – continuous wave (CW) Doppler and moving target indicator (MTI) – also exploit the difference in Doppler shift between the target and clutter in order to improve detection. Pulse-Doppler radar, however, offers several advantages relative to these modes. Compared to unmodulated CW Doppler, the pulsed mode of operation offers much better transmit–receive isolation, especially for a monostatic radar, and the potential for significantly better range estimation. Compared to MTI, the narrowband Doppler filtering offers superior 175

176

CHAPTER

5

Airborne Pulse-Doppler Radar

clutter-rejection capability, especially given the wide Doppler extent of the clutter interference that results from the high speed of the radar platform itself, a major consideration for airborne pulse-Doppler radar that is not present in surface variants. The benefits of pulse-Doppler radar are not without cost. Pulse-Doppler radar measurements are generally ambiguous in either range or velocity or both. In fact, pulseDoppler waveforms are classified based on the nature of their ambiguities: high PRF (HPRF) is unambiguous in velocity but generally highly ambiguous in range; low PRF (LPRF) is unambiguous in range but generally highly ambiguous in velocity; and medium PRF (MPRF) is moderately ambiguous in both range and velocity. This ambiguous nature results not only in uncertainty of the target’s true position or velocity but also in the folding of the clutter energy in the ambiguous dimension(s) and in the creation of range or velocity blind zones. These complications are particularly troublesome in search and acquisition modes, during which there is little or no prior information regarding target position and velocity. The pulse-Doppler radar generally circumvents these problems through the use of multiple pulse bursts at different PRFs, thereby resulting in increased radar timeline, a penalty that can be very costly when multiplied by many beam positions associated with a search pattern. The main objective of this chapter is to describe how an airborne pulse-Doppler radar isolates targets of interest from clutter and measures target range and velocity. Related topics such as pulse compression, automatic detection, target tracking, and target and clutter statistics are described in [1]. The intent is to present basic principles and concepts. Although some of the approximations used sacrifice generality and rigor for the sake of clarity, they are otherwise fairly accurate for most conditions of interest. The material in this chapter is drawn largely from [2]. Table 5.1-1 lists many of the symbols used in this chapter for subsequent reference. The remainder of the chapter is organized as follows. Section 5.2 defines the geometry and coordinate system used to describe the relative position and motion of the radar, targets, and clutter. Section 5.3 presents an approximate derivation of the Doppler shift and illustrates why airborne radars might need Doppler processing to detect targets in the presence of clutter. Section 5.4 derives expressions for the range and Doppler extents of the three main clutter components: mainlobe clutter, sidelobe clutter, and the altitude return (the vertical-incidence return). Section 5.5 shows how specific range and Doppler values of clutter map onto Earth’s surface through isorange and iso-Doppler contours. Section 5.6 presents an example scenario that illustrates the relative range-Doppler distribution of clutter and several different types of targets. Section 5.7 presents a top-level description of the operation of the pulseDoppler radar, including the transmit-waveform characteristics and the formation of range gates and Doppler filters. Section 5.8 describes the ambiguities that arise due to sampling at the PRF rate that is intrinsic to pulse-Doppler waveforms, and the resulting implications of clutter folding (aliasing) and blind zones. Section 5.9 provides an overview of the three different PRF regimes – high, medium, and low – and summarizes their relative strengths and limitations. Section 5.10 describes the HPRF mode in additional detail, including methods of measuring range and the nature of range eclipsing. Section 5.11 describes the MPRF mode in additional detail, including blind zone charts, cumulative probability of detection and false alarm, ambiguity resolution, and sidelobe blanking. Section 5.12 relates the LPRF Doppler beam sharpening mode

5.2 TABLE 5.1-1

¢

Definition of Symbols

Symbol Definition A B c d fc fD fp G h hr ht MR Nb ND Np R RC re RG Rh RU Tcpi Tp vR vr vt vU

Geometry

Symbol Definition

Area a Waveform bandwidth e Speed of light in vacuum (3 108 m/s) e0 Duty factor ev RF carrier frequency eh Doppler frequency f Pulse repetition frequency (PRF) g Antenna gain h Height above Earth’s surface y Radar aircraft height l Target height q Number of range bins q3dB Number of pulses in a burst qnull Number of Doppler bins dC Number of pulses processed dD Slant range dR Cross range dv Radius of Earth s Ground range s Range to radar horizon t Unambiguous range tC Coherent processing interval w Pulse repetition interval (PRI) x Radial velocity, or range rate z Radar aircraft velocity Dfbin Target velocity DRbin Unambiguous velocity

Azimuth angle Elevation angle Depression angle (e0 e) Aircraft climb angle Elevation angle to the radar horizon Phase Grazing angle Angle of incidence Angle relative to velocity vector Wavelength Beamwidth (general) Half-power beamwidth Null-to-null beamwidth Cross-range resolution Doppler resolution Range resolution Velocity resolution Radar cross section Backscatter coefficient Uncompressed pulse width Compressed pulse width Angular frequency Linear frequency modulation rate Angle relative to –z direction Doppler bin, or Doppler filter, width Range bin, or range gate, width

to synthetic aperture radar (SAR). Section 5.13 provides a brief summary. References are listed in Section 5.14.

5.2

GEOMETRY

5.2.1 Coordinate System The geometry illustrated in Figure 5.2-1 will be used throughout this chapter. The radar is located at the origin of a Cartesian coordinate system, whose are body-fixed with respect to the radar platform. The z-axis corresponds to the vertical direction, with negative z being the direction of acceleration due to gravity (down). The unit vectors in the x, y, and z directions are denoted by b x, b y , and bz , respectively. The notation used for ! any three-dimensional vector u is summarized in Table 5.2-1. The radar is at height hr above Earth’s surface. The point directly below the radar on Earth’s surface has coordinates (0, 0, –hr). The radar velocity is ! y þ vrzbz x þ vryb v r ¼ vrx b

(5.2-1)

177

178

CHAPTER

Airborne Pulse-Doppler Radar

5

FIGURE 5.2-1 ¢ Coordinate System.

z Scatterer → R

ψ

Airborne Radar

→ vr

Rz y

e

a

Rx

hr Ry

Earth’s Surface

x

TABLE 5.2-1 Symbol ! u u _ u ux, uy, uz

¢

Notation for a Vector in Cartesian Coordinate System Definition Three-dimensional vector of somepparameter ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u Magnitude of ! u ; i.e., u ¼ j! uj¼ ! u ! u _ Unit vector in the direction of ! u ; i.e., u ¼ ! u =u Components of ! u in the x, y, and z directions; _ i.e., ux ¼ ! u x , etc.

The position of an object, or scatterer, in the environment relative to the radar is denoted by ! x þ Ryb R ¼ Rx b y þ Rzbz

(5.2-2)

The object’s position can also be described in terms of the slant range R, azimuth angle a, and elevation angle e, relative to the radar. The slant range to the object is ffi ! qffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi ! ! qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (5.2-3) R ¼ R ¼ R R ¼ R2x þ R2y þ R2z The azimuth angle is the clockwise angle between the y-axis and the projection of ! ! R onto the xy plane. The elevation angle is the angle of R relative to the xy plane, defined as positive for scatterers above the xy plane and negative for scatterers below it, such as terrain clutter. It is often convenient to use the depression angle e0 , defined as the negative of the elevation angle, to provide an implicitly positive quantity for directions ! below the xy plane. The x-, y-, and z-components of R are related to a, e, and e0 through Rx ¼ R cosðeÞsinðaÞ ¼ R cosðe0 ÞsinðaÞ

(5.2-4)

Ry ¼ R cosðeÞcosðaÞ ¼ R cosðe0 ÞcosðaÞ

(5.2-5)

Rz ¼ R sinðeÞ ¼ R sinðe0 Þ

(5.2-6)

5.2

Geometry

5.2.2 Angle Relative to the Radar Velocity Vector The angle, y, between the radar velocity vector ! v r and the line-of-sight (LOS) direction ! of a scatterer at point R is found through the relationship ! ! v r R ¼ vr R cosðyÞ

(5.2-7)

Using the preceding expressions for Rx, Ry, and Rz, from Equations 5.2-4 through 5.2-6, we can express y in terms of the velocity vector and the azimuth and elevation angles: ! ! v r R vrx Rx þ vry Ry þ vrz Rz cosðyÞ ¼ ¼ vr R vr R vrx cosðeÞ sinðaÞ þ vry cosðeÞ cosðaÞ þ vrz sinðeÞ ¼ vr

(5.2-8)

Without loss of generality, we define the positive y direction to be that of the horizontal component of radar velocity, that is, vrx ¼ 0. The preceding expression becomes cosðyÞ ¼

vry cosðeÞ cosðaÞ þ vrz sinðeÞ vr

(5.2-9)

The radar aircraft climb angle, ev , is related to the velocity x-, y-, and z-components through ev ¼ tan1 ðvrz =vry Þ ¼ sin1 ðvrz =vr Þ ¼ cos1 ðvry =vr Þ

(5.2-10)

We can then alternatively express the angle y through cosðyÞ ¼ cosðev Þ cosðeÞ cosðaÞ þ sinðev Þ sinðeÞ

(5.2-11)

For the simple case in which the radar is flying entirely in the horizontal direction (en ¼ 0), the expression for angle relative to the radar velocity reduces to cosðyÞjev ¼0 ¼ cosðeÞ cosðaÞ

(5.2-12)

For the case in which there is vertical motion, but we constrain the position vector to be in the yz plane, that is, a ¼ 0 or a ¼ p, the expression becomes cosðyÞja¼0 ¼ cosðev Þ cosðeÞ þ sinðev Þ sinðeÞ ¼ cosðev eÞ

(5.2-13)

cosðyÞja¼p ¼ cosðev Þ cosðeÞ þ sinðev Þ sinðeÞ ¼ cosðev þ eÞ

(5.2-14)

where we use the trigonometric identity cos (a þ b) ¼ cos(a) cos(b) – sin(a) sin(b).

5.2.3 Range and Elevation Angle to a Point on the Earth Surface 5.2.3.1 Flat-Earth Approximation We can approximate the slant range R and elevation angle e to a point on Earth’s surface by assuming a flat earth as shown in Figure 5.2-2. With this simplified model, the range and elevation (or depression) angle for a given radar height are related through hr ¼ R sinðeÞ ¼ R sinðe0 Þ

(5.2-15)

179

180 FIGURE 5.2-2 ¢ Geometry for Flat-Earth Approximation.

CHAPTER

Z

5

Airborne Pulse-Doppler Radar Airborne Radar ε′ ζ

hr = R sin(ε′)

R

hr

RG = R cos(ε′) η γ

Earth’s Surface

RG

Scatterer

The grazing angle, g, is the LOS angle relative to Earth’s surface at the reflection point. The angle of incidence, h, is the LOS angle relative to the normal to Earth’s surface at the reflection point. For the flat-earth model, these are given by g ¼ e0

(5.2-16)

h ¼ p=2 e0

(5.2-17)

! The projection of R onto the flat-earth surface is referred to here as the ground range, RG, given by RG ¼ Rcos(e0 ). The flat-earth assumption is sufficiently accurate to convey most of the important concepts relevant to airborne pulse-Doppler radar and will be implicitly used in this chapter unless otherwise noted.

5.2.3.2 Spherical-Earth Approximation One limitation of the flat-earth approximation is that it does not provide a means for estimating the range and elevation angle to the radar horizon. For this purpose, we employ a spherical-earth model as shown in Figure 5.2-3, where re is the radius of Earth, and the angle z is the LOS angle relative to the negative z-axis. From the law of cosines and the fact that z ¼ p=2 e0 (implying cosðzÞ ¼ sinðe0 Þ), we can relate range, height, and depression angle through R2 þ ðre þ hr Þ2 2Rðre þ hr Þ sin e0 re2 ¼ 0

(5.2-18)

The depression angle in terms of slant range and height can therefore be found from hr R2 h2r 0 sinðe Þ ¼ 1þ (5.2-19) R 2hr ðre þ hr Þ The maximum range of any point on Earth’s surface visible to the radar occurs at the radar horizon where the LOS angle is tangential to Earth’s surface, that is, h ¼ p/2 in Figure 5.2-3. From the Pythagorean theorem, the range to the radar horizon, Rh, is qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffi Rh ¼ ðre þ hr Þ2 re2 ¼ 2re hr þ h2r ffi 2re hr (5.2-20) pffiffiffiffiffiffiffiffiffiffi The approximation of Rh ffi 2re hr is valid for hr eh sinðev Þ

(5.4-11)

The maximum and minimum Doppler shifts over all ranges are therefore 2vr cosðev eh Þ ½ev eh l 2vr ½ev < eh ¼ l 2vr cosðev þ eh Þ ½ev eh fD;SLCmin ¼ l 2vr ½ev > eh ¼ l

fD;SLCmax ¼

(5.4-12)

(5.4-13)

Figure 5.4-4 illustrates the notional shape of the sidelobe clutter range and Doppler bounds assuming purely horizontal motion (ev ¼ 0). Table 5.4-2 lists the sidelobe clutter Doppler extent as a function of range for different ranges and heights, with vry ¼ 300 m/s, vrz ¼ 0, and the radar frequency equal to 10 GHz (l ¼ 3 cm). The Doppler extent for each height rapidly approaches 40 kHz ( 20 kHz), corresponding to the upper and lower bounds of 2vr =l.

189

190

CHAPTER

5

Airborne Pulse-Doppler Radar R

FIGURE 5.4-4 ¢ Sidelobe Clutter Range-Doppler Distribution.

R = Rh SLC Boundary

ΔfD,SLC = ±

2vr λ

h2 1– r2 R

R = hr –

TABLE 5.4-2

¢

2vr

+

λ

2vr

fD

λ

Example of Sidelobe Clutter Doppler Extent versus Range and Altitude Doppler Extent (Hz) per Aircraft Height hr

R (m)

hr ¼ 500 m (Rh ¼ 79.9 km)

hr ¼ 1,000 m (Rh ¼ 112.9 km)

hr ¼ 2,000 m (Rh ¼ 159.7 km)

hr ¼ 3,000 m (Rh ¼ 195.6 km)

600 1,100 2,100 3,100 5,000 10,000 20,000 50,000

22,111 35,629 38,850 39,476 39,799 39,950 39,987 39,998

0 16,664 35,174 37,862 39,192 39,799 39,950 39,992

0 0 12,196 30,562 36,661 39,192 39,799 39,968

0 0 0 10,078 32,000 38,158 39,547 39,928

(vry ¼ 300 m/s; vrz ¼ 0; fc ¼ 10 GHz)

*

The power in the sidelobe clutter spectrum at a given range is not uniformly distributed across all Doppler filters, being, of course, highly dependent on the radar antenna sidelobe pattern. We also note that the physical area of a clutter patch contained in a given Doppler bin increases in the vicinities of a ¼ 0 and a ¼ p. To illustrate this, consider the variation in Doppler frequency DfD resulting from a small variation in azimuth Da: @fD @ 2vr 2vr Da ¼ cosðeÞ cosðaÞ Da ¼ cosðeÞ sinðaÞDa (5.4-14) DfD ffi @a l @a l Rearranging terms, we see that the azimuth extent, and therefore the physical area, contained within a fixed Doppler interval is greatest in the regions near a ¼ 0 and a ¼ p: Da l ffi (5.4-15) Df 2v cosðeÞ sinðaÞ D r

5.4.3 Mainlobe Clutter Range and Doppler Extent 5.4.3.1 Mainlobe Clutter-Range Extent The mainlobe clutter extends in range from the intersection of the lower edge of the beam with the terrain to either the corresponding intersection of the upper edge of the

Range and Doppler Distribution of Clutter

5.4 Radar

FIGURE 5.4-5 ¢ Mainlobe Clutter Range Extent.

′ RMLC,min R t

θnull 2

hr

∆h = hr – ht

θnull 2 RSLC,min

ht Target

beam, if the entire beam illuminates the terrain, or the radar horizon, if the upper portion of the beam does not illuminate the terrain. Figure 5.4-5 illustrates the former condition. The minimum mainlobe clutter range is RMLC;min ¼

hr sinðe0s þ qnull =2Þ

(5.4-16)

The maximum mainlobe clutter range is 0 hr ðes qnull =2Þ e0h qnull =2Þ pffiffiffiffiffiffiffiffiffiffiffi 0 ¼ Rh ffi 2re hr ðes qnull =2Þ < e0h

RMLC;max ¼

sinðe0s

(5.4-17)

where from Equation 5.2-15, e0h is the depression angle to the radar horizon, e0h ¼ sin1 ðhr =Rh Þ. The mainlobe clutter range extent is therefore DRMLC ¼

2hr sinðqnull =2Þ cosðe0s Þ sin2 ðe0s Þ sin2 ðqnull =2Þ

pffiffiffiffiffiffiffiffiffiffi ¼ 2re hr

191

0 ðes qnull =2Þ e0h

hr 0 sinðes þ qnull =2Þ

0 ðes qnull =2Þ < e0h

(5.4-18)

(The form of the denominator in the preceding first equation is obtained by using the substitution cos2 ðqnull =2Þ ¼ 1 sin2 ðqnull =2Þ.) As an example, consider a radar at an altitude of hr ¼ 3,000 m, with a null-to-null beamwidth of qnull ¼ 7.5 degrees, and an elevation scan angle of es ¼ –15 degrees. The range extent for this case (the entire mainbeam illuminates the terrain) is approximately 6,044 m, extending from 9,333 m to 15,377 m.

5.4.3.2 Mainlobe Clutter Doppler Extent The Doppler shift of mainlobe clutter at the center of the mainbeam for a specific antenna scan angle ys relative to the velocity vector is fD;MLCctr ¼

2vr cosðys Þ l

To a first approximation, the mainlobe clutter Doppler extent is 2vr cosðys qnull =2Þ cosðys þ qnull =2Þ DfD;MLC ffi l

(5.4-19)

(5.4-20)

192

CHAPTER

5

Airborne Pulse-Doppler Radar

assuming a circular beam and ys > qnull =2. Applying the identity cos(a þ b) ¼ cos(a) cos(b) – sin(a) sin(b) and using the approximation sinðqnull =2Þ ffi qnull =2 for qnull =2 hr) encompasses a very large area on Earth’s surface whose return may be much greater than targets of interest at longer range [3]. Although we treat it as a separate component to simplify the discussion, in fact it is generally part of a continuum of strong sidelobe returns that may occur along the elevation principle sidelobe plane between the mainbeam and the nadir direction (below the aircraft). Figure 5.4-6 illustrates the geometry for the altitude return, assuming the flat-earth model. The ground range, RG, is related to slant range through R2G ¼ R2 h2r and to DR

Radar x

z

FIGURE 5.4-6 ¢ Geometry of Altitude Return.

y ε′ R = hr+∆R

Ac = πR2G = 2πhr ∆R σc = σoAc

+

π∆R2

hr

η

RG = R2−h2r

194

CHAPTER

5

Airborne Pulse-Doppler Radar

through R2G ¼ 2DRhr þ DR2 . The area on Earth’s surface encompassed by the range extent DR is 1 Ac ¼ pR2G ¼ 2phr DR 1 þ DR=hr (5.4-28) 2

For DR/hr

(5.10-1)

4vr 2vt;max þ l l

(5.10-2)

Figure 5.10-2 illustrates the Doppler spectrum of the example scenario, using a 250-kHz PRF and a 1.5-ms pulse width. The highest magnitude Doppler shift in the example scenario is approximately 33 kHz for Target 1; the total Doppler extent of the sidelobe clutter is approximately 40 kHz; the PRF must therefore be greater than 73 kHz in order ensure that Target 1 is in the clutter-free region, a condition that is easily met by the selected value of 250 kHz. The unambiguous velocity for the 250-kHz PRF is 3,750 m/s. The unambiguous range is 600 m, indicating that all uneclipsed range returns are folded into a 600-m interval. The other airborne targets (2, 3, and 4) are all in the sidelobe clutter region. Detection of these targets in HPRF mode is difficult because of the high degree of range folding of the clutter in each target’s Doppler filter. From Equation 5.4-8 we can infer

FIGURE 5.10-1 ¢ Notional HPRF Doppler Spectrum.

1/τ PRF

PRF MLC

sinc(fDτ)

ALT

Clutter-Free Region

SLC

Target (–fp–

2vr –f 2vr p (–fp+ λ ) λ )

2vr λ

+

2vr λ

(fp–

2vr λ )

fp

(fp+

2vr λ )

Doppler

FIGURE 5.10-2 ¢ HPRF Doppler Spectrum of the Example Scenario for a 250-kHz PRF.

PRF Target 4

–20

–4.7

Target 3

9.4

Target 5 (15.5 kHz) Target 6 (17.1 kHz) Target 2 (18.8 kHz) Target 1 20

230 33 Doppler (kHz)

250

270

283

5.10

High PRF Mode

that a Doppler filter centered at frequency fn may contain sidelobe clutter returns from ranges as close as hr Rmin;fn ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi 1

fn l 2vr cosðev Þ

2

(5.10-3)

where ev is the aircraft climb angle. Consider, for example, the slow outbound aircraft, Target 3, in the example scenario. The radar is closing in on the target in a tail-chase type geometry. From Table 5.6-2, the range of Target 3 is 63.7 km, and the Doppler frequency is 9.4 kHz. The Doppler filter that contains this target therefore also includes sidelobe clutter from ranges as close as Rmin;fn ¼ 3.4 km, a factor of 18.7 times closer than the target. Since received power is proportional to R4, this corresponds to a 51-dB disadvantage for the target, offsetting much of the two-way sidelobe antenna pattern attenuation that works in favor of the mainlobe target versus the sidelobe clutter. The target must compete with clutter returns from many ambiguous or folded ranges between the preceding minimum range and the radar horizon. With a 600-m unambiguous range, there are approximately 320 ambiguous range cells between Rmin;fn and the radar horizon, of which more than 100 are at ranges closer than Target 3. (Each range cell actually includes two clutter patches – one at positive azimuth and one at negative azimuth.) The overall clutter interference is the sum of the contributions from each rangefolded clutter patch within the Doppler filter. Associated with each patch k is its own physical area Ac, backscatter coefficient s , two-way antenna pattern gain GtGr (where Gt and Gr are the transmit and receive antenna gains, respectively), and rangedependence R4 c . The impact of the folded clutter on target detection is assessed through the resulting signal-to-clutter ratio: S ffi C

st R4t

! X Ac ðkÞs ðkÞGt ðkÞGr ðkÞ k

R4c ðkÞ

(5.10-4)

5.10.2 Linear FM Ranging HPRF 5.10.2.1 Linear FM Ranging Concept One method of obtaining range information with an HPRF waveform is to use LFM ranging. This is referred to as range-while-search mode in the F-15 radar. The radar linearly varies the RF carrier frequency by a relatively small amount over the course of the transmitted pulse burst. The LFM is generated in one of the local oscillators that is common to both transmit and receive paths. Without any LFM, the frequency of a signal received from an object in the environment differs from the radar’s instantaneous transmitter frequency by an amount equal to the object’s Doppler shift fD;vR ¼2vR =l. With the application of LFM, an additional frequency difference, fD,LFM, is imparted to the object due to the change in transmitter (and receiver) frequency between the time the signal was originally transmitted and the time the reflection is received. Let us represent the transmitter (and receiver) carrier frequency by fc(t) ¼ fc,0 þ xt, where fc,0 is the frequency at the beginning of the burst (t ¼ 0), and x is the LFM rate

231

232

CHAPTER

5

Airborne Pulse-Doppler Radar

during the burst, expressed in units of frequency per unit time. The additional frequency shift of an object at range R that is induced by the LFM is therefore fD;LFM ¼ fc ðt 2R=cÞfc ðtÞ ¼ ½fc;0 þðxt 2Rx=cÞ½fc;0 þxt ¼ 2xR=c

(5.10-5)

The total frequency shift of a return from range R is fD ¼ fD;vR þ fD;LFM ¼

2vR 2xR c l

(5.10-6)

Two bursts at LFM rates x1 and x2 produce two different shifts, fD1 and fD2 . The frequency difference between fD1 and fD2 allows the range of the object to be determined: 2vR 2x2 R 2vR 2x1 R 2ðx x ÞR ¼ 1 2 (5.10-7) Df12 fD2 fD 1 ¼ c c c l l )R¼

cDf12 2ðx1 x2 Þ

(5.10-8)

The concept of LFM ranging is illustrated in Figure 5.10-3. The figure illustrates three bursts: the first without LFM, the second with a positive LFM, and the third with a negative LFM. In general, more than two LFM rates are desirable to avoid ambiguities that arise due to multiple targets in the environment. Two objects separated by DR can be resolved in range if the frequency displacement experienced by each differs by at least one Doppler filter: jDf12 ðRÞ Df12 ðR þ DRÞj Dfbin

(5.10-9)

where Dfbin is the Doppler filter width. If we approximate the Doppler filter width as 1/Tcpi, where Tcpi is the CPI time, we can estimate the achievable range resolution, dR: c Df12 ðRÞ Df12 ðR þ DRÞ cDfbin DR ¼

(5.10-10) 2ðx1 x2 Þ 2jx1 x2 j FIGURE 5.10-3 ¢ Concept of LFM Ranging.

∆t = 2R/c

fD2 = –

2vR −x2Δt λ

Frequency

fD3 = –

fD1 = −

Target Return

2vR λ x2 < 0

x3 < 0

x1 =0

Tb

2vR −x3Δt λ

2Tb

Tcpi

Tcpi

Burst 1

Burst 2

Transmitter (& Receiver)

3Tb Tcpi Burst 3

Time

5.10

) dR ¼

High PRF Mode

cDfbin c ¼ 2jx1 x2 j 2jx1 x2 jTcpi

233

(5.10-11)

The product (|x1 – x2|Tcpi) in the denominator is actually the bandwidth, B, separating the two returns separated by range dR. This leads to the familiar result for range resolution in terms of signal bandwidth: dR ¼ c/2B. Suppose that an HPRF mode employs a 10-MHz/s LFM rate and a 5-ms CPI. The resulting bandwidth is 50 kHz, corresponding to a range resolution of approximately 3 km. The requirement for multiple CPIs potentially impacts either radar timeline or sensitivity. Either the total time on target must increase to preserve the same single-CPI SNR, thereby impacting search frame time, or the CPI time per burst must decrease, thereby impacting probability of detection.

5.10.2.2 Clutter Spreading Due to LFM Ranging One consequence of using LFM ranging in the HPRF mode is that the clutter is spread over a wider portion of the Doppler spectrum. In Section 5.4.3.2, we found that the Doppler shift of sidelobe clutter at a given range, R, falls within the bounds of rffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2vr h2 fD;SLCmin;R ¼ 1 r2 (5.10-12) l R rffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2vr h2 1 r2 (5.10-13) fD;SLCmax;R ¼ þ l R assuming horizontal flight. The square-root term is just the cosine of the depression angle under the flat-earth approximation. With LFM ranging, each of the preceding terms is offset by the addition of

2xR c

to account for the range-dependent frequency

shift induced by the LFM. This imparts a small frequency shift to close-range sidelobe clutter and a large frequency shift to long-range sidelobe clutter. The result is therefore not simply a shifting of the entire spectrum, but both a shifting and a spreading of the spectrum. This spreading cuts directly into the clutter-free region of the HPRF spectrum. Figures 5.10-4a and 5.10-4b illustrate the spreading effect for the example scenario, using LFM slopes of þ 10 MHz/s and –10 MHz/s, respectively. The Doppler extents of

Without LFM: With LFM:

With LFM:

200

200

150

150 Range (km)

Range (km)

FIGURE 5.10-4 ¢ Spreading of Clutter Spectrum Due to LFM Ranging in HPRF.

Without LFM:

100 50

MLC

100 50

SLC

0 –40 –30 –20 –10 0 10 20 30 40 Doppler (kHz)

0 –40 –30 –20 –10 0 10 20 30 40 Doppler (kHz)

(a) +10 MHz/s LFM.

(b) –10 MHz/s LFM.

234

CHAPTER TABLE 5.10-1

5 ¢

Airborne Pulse-Doppler Radar Example of Clutter Doppler Spread from LFM Ranging HPRF*

Without LFM

fD,min fD,min fD,max Df (kHz) (kHz) (kHz) (kHz) SLC MLC

–20.0 þ13.2

þ20.0 40.0 þ15.1 1.8

With þ10 MHz/s LFM

With –10 MHz/s LFM

–18.6 15.8

fD,max Df (kHz) (kHz)

Spread fD,min fD,max Df Spread (kHz) (kHz) (kHz) (kHz) (kHz)

33.4 28.0

11.0 10.4

51.0 12.2

–33.4 0.3

þ18.6 51.0 12.3 12.0

11.0 10.2

fc ¼ 10 GHz, vr ¼ 300 m/s, hr ¼ 3,000 m.

*

sidelobe clutter and mainlobe clutter are listed in Table 5.10-1 for the two LFM slopes as well as for the non-LFM case. The selected LFM rates cause the sidelobe clutter to spread by 11 kHz, or by approximately 27 percent, relative to the non-LFM extent of 40 kHz. The mainlobe clutter spreads by 10.2 or 10.4 kHz, or approximately 560 percent relative to its non-LFM extent of 1.8 kHz.

5.10.3 Range-Gated HPRF Range-gated HPRF mode allows the PRI to be subdivided into range gates, providing improved range resolution and range measurement relative to conventional HPRF. The range gates may be achieved through lower duty cycle (shorter pulse widths or longer PRI or both) and low time-bandwidth product pulse compression. The range gates provide a highly ambiguous range measurement that can be resolved through a combination of LFM ranging, as previously described, and ambiguity-resolution methods similar to those employed in medium PRF modes, described in Section 5.11. The improved range resolution has two benefits. First, it improves situation awareness by allowing detection of multiple targets flying at the same radial velocity but at different ranges. Second, it allows the clutter within a Doppler filter to be subdivided in the range dimension. This may improve performance against targets with Doppler frequencies within the bounds of sidelobe clutter. Strong returns that are highly localized in range, such as the altitude return and close-range sidelobe discretes, can be isolated from the target through the range resolution.

5.10.4 HPRF Range Eclipsing One disadvantage of high duty cycle HPRF waveforms is the range eclipsing, illustrated in Figure 5.10-5. The probability is high (approximately twice the duty factor) that the return signal from a target will be at least 50 percent eclipsed at any given time during search mode (prior to initial detection). For a given PRF, a target of radial velocity vR will fly in and out of the eclipsing condition over a period of Teclipse ¼

RU cTp c ¼ ¼ jvR j 2jvR j 2fp jvR j

(5.10-14)

For example, if the radial velocity is vR ¼ –600 m/s and the PRF is fp ¼ 250 kHz (Tp ¼ 4 ms), then the eclipsing period is Teclipse ¼ (3 108 m/s ) / (2 250 103 Hz 600 m/s) ¼ 1.0 s. Rather than merely waiting for the target to fly out of an eclipsed region, the radar can actively change the eclipsing condition by varying the PRF. The amount of PRF

5.11

τ

Medium PRF Mode

Tp

FIGURE 5.10-5 HPRF Target Eclipsing.

Transmit τblank Receiver Blanking

Uneclipsed Return Partially Eclipsed Return Fully Eclipsed Return

variation needed to move an eclipsed target into an uneclipsed region is relatively small. Suppose a target at range Rt is fully eclipsed with PRI Tp1. The total number of PRIs that correspond to range Rt for this PRI is k1 ¼

Rt 2Rt 2Rt fp1 ¼ ¼ RU1 cTp1 c

(5.10-15)

The nearest PRI that would also produce an integer multiple of PRI intervals to Rt is Tp2: k 2 ¼ k1 þ 1 ¼

Rt 2Rt 2Rt fp2 ¼ ¼ RU 2 cTp2 c

(5.10-16)

The relative change in PRF is therefore fp2 fp1 c RU1 ¼ ¼ fp1 Rt 2Rt fp1

(5.10-17)

For a 30-km target and a 250-kHz PRF (RU1 ¼ 600 m), the relative PRF change needed to move the target through one complete eclipsing cycle is (600/30,000) ¼ 2 percent. Thus, a 1-percent change, or approximately 2.5 kHz, would be sufficient to move the target half a PRI into the clear.

5.11

235

MEDIUM PRF MODE

Medium PRF mode provides good capability under a wide range of conditions. Although MPRF waveforms are ambiguous in both range and Doppler, the ambiguities are sufficiently few that they can be resolved through the use of multiple PRFs, allowing both range and velocity measurements to be made of the target. The radar divides the PRI into many range gates, often using pulse compression to achieve good range resolution while maintaining reasonably high duty cycle. For example, a radar that uses

¢

236

CHAPTER

5

Airborne Pulse-Doppler Radar

a 10-kHz PRF (100-ms PRI), a 6.5-ms pulse width, and a 13-bit Barker code pulse compression, has a 6.5-percent duty cycle and approximately 200 range gates over the PRI (13 100/6.5 ¼ 200), with a range resolution of approximately 75 m (tc ¼ 6.5 ms/ 13). A 32-pulse CPI at this PRF would be 3.2-ms long and would allow the 10-kHz folded Doppler spectrum to be divided into 32 Doppler filters (assuming FFT-type processing). Its main disadvantage relative to HPRF is that it does not provide a clutterfree region of the Doppler spectrum for optimum detection of high closing-rate targets.

5.11.1 Examples of Different PRFs

PRF 100 90 MLC 5 80 4 70 3 PRI ×c/2 60 2 SLC 50 1 40 30 20 ALT 10 6 0 –40 –30 –20 –10 0 10 20 30 40 Doppler (kHz) (a) Ambiguous Regions.

30

Range (km)

2

SLC

5

25

Range (km)

FIGURE 5.11-2 ¢ Range-Doppler Folding of Example Scenario for 10-kHz MPRF.

PRF 100 90 MLC 5 80 4 PRI× c/2 70 3 60 2 SLC 50 1 40 30 20 ALT 10 6 0 –40 –30 –20 –10 0 10 20 30 40 Doppler (kHz) (a) Ambiguous Regions.

Range (km)

FIGURE 5.11-1 ¢ Range-Doppler Folding of Example Scenario for 5-kHz MPRF.

Range (km)

Figures 5.11-1, 5.11-2, and 5.11-3 illustrate the range-Doppler spectrum for three different PRFs in the MPRF range: 5 kHz, 10 kHz, and 18 kHz. The figures show the range-Doppler map of the example scenario divided into the corresponding ambiguous range and Doppler intervals (a) and the resultant folded spectrum (b). The range extent of both mainlobe clutter and sidelobe clutter exceeds the unambiguous range for each PRF; therefore, mainlobe clutter and sidelobe clutter are present at all ranges in the folded spectrum. The 40-kHz sidelobe clutter Doppler extent exceeds each PRF, and therefore sidelobe clutter is present at all Doppler frequencies in the folded spectrum. The 5-kHz PRF places Targets 2 and 3 in mainlobe clutter, indicating they would be in a velocity blind zone for this PRF. The 10-kHz PRF results in all targets being in the clear.

1

20

MLC

15 4 10 6

ALT

5

3

0 0

15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0

1

2 3 4 Doppler (kHz) (b) Folded Range-Doppler Image.

5

MLC SLC

2

4 5 6 1

ALT

1

2

3

3

4 5 6 7 Doppler (kHz)

8

(b) Folded Range-Doppler Image.

9

10

5.11 PRF

8

100 MLC

80 70 Range (km)

5 PRI ×c/2

3

60 50

MLC

2

6

4

SLC

2 1

40 30

3 6

5 4 4

ALT

3 2

20 10

5 1

ALT 6

0 –40 –30 –20 –10 0 10 Doppler (kHz)

20

0 30

(a) Ambiguous Regions.

40

2

4

6

8 10 12 Doppler (kHz)

14

16

237 FIGURE 5.11-3 ¢ Range-Doppler Folding of Example Scenario for 18-kHz MPRF.

1

SLC

7

Range (km)

90

Medium PRF Mode

18

(b) Folded Range-Doppler Image.

The 18-kHz PRF puts Targets 1 and 4 at the edges of mainlobe clutter. The altitude return clutter spectrum is nominally 6 kHz, and therefore does not occupy the entire Doppler extent for this PRF. Notice that there is no way of distinguishing the various target types, including ground mover and sidelobe discrete, based on their location in the folded range-Doppler image for any given PRF. Thus, for a single CPI, they are all potentially valid target detections.

5.11.2 Blind-Zone Charts Medium PRFs have blind zones in both range and Doppler that impact target detection in search mode. (In contrast, during track mode, the PRF can be chosen to keep the target in the clear because the target range and velocity have already been determined.) One way of illustrating the effect of waveform selection on target visibility is with a blind-zone chart. Figure 5.11-4 illustrates a blind-zone plot for a 10-kHz PRF, a 10-ms pulse width, and a 2-kHz mainlobe clutter Doppler extent (Doppler blind zone), with the mainlobe clutter tuned to zero Doppler. The left and right vertical scales are in time delay and range, respectively, and the top and bottom horizontal scales are in velocity and Doppler shift, respectively, assuming a 10-GHz frequency. The chart covers the region of 0–60 km range (0–400 ms time) and 0–600 m/s velocity (0–40 kHz Doppler). The range blind zones occur at integer multiples of the unambiguous range (cTp/2), 15 km. Each range blind zone extends for at least the equivalent range of the transmit pulse width, ct/2, or 1.5 km. The velocity blind zones occur at integer multiples of the unambiguous velocity (lfp/2), 150 m/s. Each velocity blind zone extends for half the velocity extent of the mainlobe clutter ( lDfMLC/4), or 15 m/s. A radar that employs this waveform in a search mode is blind to approximately 28 percent of the potential targets in the environment [1 – (90 ms 8 kHz)/(100 ms 10 kHz) ¼ 1 – 0.72 ¼ 0.28]. Suppose the radar maintains the same pulse width but switches to a slightly different PRF of 11 kHz, as shown in Figure 5.11-5. Although the range blind zone at the minimum range and the velocity blind zone at zero Doppler remain in the same positions, the 11-kHz PRF produces blind zones that are otherwise at generally different locations than for the 10-kHz PRF. (The 10-kHz blind zones are outlined with the dashed lines for comparison.) A third PRF of 12 kHz would yield yet a different set of blind zones.

238 FIGURE 5.11-4 ¢ Blind-Zone Chart for 10-kHz PRF (fc ¼ 10 GHz, t ¼ 10 ms, DfD,MLC ¼ 2 kHz, Mainlobe Clutter Tuned to 0).

CHAPTER

5

Airborne Pulse-Doppler Radar

vU 0

Velocity (m/s)

100

200

300

400

500

600 60

400

τ

350 50 300 40

Delay (μs)

30

200

Range (km)

250

150 20 100 10

PRI 50

0 0

5

10

15

20

25

30

35

100

200

40

ΔfD,MLC

Doppler (kHz)

PRF

FIGURE 5.11-5 ¢ Blind-Zone Chart for 11-kHz PRF (fc ¼ 10 GHz, t ¼ 10 ms, DfD,MLC ¼ 2 kHz, Mainlobe Clutter Tuned to 0).

RU

Velocity (m/s) 300 400

500

400

600 60 10 kHz

350

50

40

250 200

30

150

Range (km)

Delay (μs)

300

20

100 10

50 0

0 0

5

10

15 20 25 Doppler (kHz)

30

35

40

Figure 5.11-6 depicts a blind-zone chart illustrating the regions for which a target is in the clear for none, one, two, or all three of the above PRFs. The darkest regions are those for which the target is in a blind zone for all three PRFs. The lightest regions represent those for which the target is in the clear for all three PRFs. The probability that a target is in the clear for at least one of the three PRFs is much greater than it is for any one PRF alone.

Medium PRF Mode

5.11 0 clear 2 clear

100

200

Velocity (m/s) 300

1 clear 3 clear

400

500

400

FIGURE 5.11-6 ¢ Composite BlindZone Chart for 10-, 11-, and 12-kHz PRFs (fc ¼ 10 GHz, t ¼ 10 ms, DfD,MLC ¼ 2 kHz, Mainlobe Clutter Tuned to 0).

600 60

350

50

300

Delay (μs)

200

30

150

Range (km)

40

250

20

100 10

50 0

0 0

5

10

15 20 25 Doppler (kHz)

30

35

239

40

5.11.3 Cumulative Probability of Detection and False Alarm A radar using MPRF during a search mode may cycle through several PRFs in order to overcome blind zones and to resolve range ambiguities. At least two PRFs are needed to resolve the range ambiguity of a detected target (see Section 5.11.4). Since it is not guaranteed that the target will be in the clear for every PRF that is selected, the radar must generally use more PRFs than the minimum number needed for ambiguity resolution. Therefore, three might be considered a minimum number of PRFs. Increasing the number of PRFs increases the probability of the target being in the clear for the minimum number needed for detection. One approach to PRF variation is referred to as an M-out-of-N criterion, whereby M detections are required out of N PRFs (M < N) in order to consider the detection valid. Consider the example of M ¼ 3 and N ¼ 8, in which the target must be detected in at least three out of the eight PRFs used. We are interested in the cumulative probability of detection Pd and cumulative probability of false alarm Pfa. Note that the cumulative Pd discussed here is different from the cumulative Pd resulting from successive scans in a search mode. In this case, Pd is the probability that a target will be detected in at least three PRFs, and Pfa is the probability that a false alarm will occur at the same range and velocity in at least three PRFs. Let pd,1 be the single-CPI probability of detection for each PRF for which the target is in the clear. Let pfa,1 be the single-CPI probability of false alarm for each PRF. (In practice, the single-CPI probability of detection and probability of false alarm may vary with different PRFs, depending on duty factor and CPI time. For the sake of discussion, we assume that the total energy per CPI is constant for each PRF. This can be achieved by using the same pulse width and total number of pulses per CPI, for example.) In the case of a pulse-Doppler radar, there is one detection opportunity for each range-Doppler cell.

240

CHAPTER

5

Airborne Pulse-Doppler Radar

Suppose that the target is in the clear for Nc of the N PRFs (Nc N). If Nc is less than M, the minimum number required for a valid detection, then by definition the cumulative probability of detection Pd is zero. If Nc M, there is a nonzero probability that the target will be detected in at least M out of the Nc clear PRFs given by Pd ¼

Nc X m¼M

Nc ! pm ð1 pd;1 ÞNc m m!ðNc mÞ! d;1

(5.11-1)

The preceding index m indicates the number of PRFs out of the Nc clear PRFs in which detections occur. The summation adds the individual probabilities of getting m detections out of the Nc PRFs: first m ¼ M PRFs (the minimum), then m ¼ M þ 1, and Nc m so on until reaching the maximum possible of m ¼ Nc. The product pm is d;1 ð1 pd;1 Þ the probability of detection for a given set of m PRFs out of Nc, that is, the probability that detections occur on m PRFs, pm d;1 , multiplied by the probability that a detection does – m) PRFs, ð1 pd;1 ÞNc m . not occur on the remaining (N c Nc Nc ! The expression . It is the binomial coefficient, denoted by m!ðNc mÞ! m represents the number of possible combinations of Nc PRFs taken m at a time [6]. A common approach to deriving this term is first to consider the total number of ways of forming an ordered sequence of m PRFs out of a set of Nc distinct PRFs without repeating a given PRF. Any of the Nc PRFs can be chosen for the first in the sequence. The pool from which to select the second is reduced by the first choice to Nc – 1 PRFs. This continues until reaching the mth PRF of the sequence, by which time the selection pool is reduced to a set of Nc – m þ 1 remaining PRFs. Thus, the total number of ordered sequences, or permutations, PNc :m , is PNc :m ¼ ðNc ÞðNc 1Þ:::ðNc m þ 1Þ ¼

ðNc ÞðNc 1Þ:::ðNc m þ 1ÞðNc mÞðNc m 1Þ:::ð2Þð1Þ ðNc mÞðNc m 1Þ:::ð2Þð1Þ

¼

Nc ! ðNc mÞ!

(5.11-2)

We are not interested in the specific order of selection of a given set of PRFs, however, so we must divide the preceding quantity by the number of different ways of ordering each unique set of m PRFs. Using the same approach, there are m choices for the first in the sequence, m – 1 choices for the second, and so on, until there is only one choice for the mth in the sequence (the remaining m – 1 having already been selected). The total number of ways of ordering a set of m PRFs is thus m factorial: m! ¼ m(m – 1) (m – 2) . . . (1). The total number of unique sets of m PRFs out of a set of Nc PRFs is therefore the binomial coefficient: PNc :m Nc ! Nc ¼ ¼ (5.11-3) m m!ðNc mÞ! m! We can also find an expression for the cumulative probability of false alarm using an analogous approach. In this case, however, we are not restricted to selecting from a set of Nc clear PRFs but instead may select from the entire set of N PRFs because a false

Medium PRF Mode

5.11

241

alarm is equally likely for both clear and blind PRFs. Thus, the cumulative probability of false alarm in at least M out of N PRFs is N X N! N m Pfa ¼ (5.11-4) pm fa;1 ð1 pfa;1 Þ m!ðN mÞ! m¼M Let us continue with the preceding example of requiring M ¼ 3 detections out of N ¼ 8 total PRFs and assume that the probability of detection on a single clear PRF is pd,1 ¼ 0.5 and the probability of false alarm on any single PRF is pfa,1 ¼ 10–3. Table 5.11-1 shows the resulting values for Pd and Pfa. Each row represents the contribution to the cumulative probability from the condition of having exactly m detections. The probability of detection is a function of the number of clear PRFs listed in columns Nc ¼ 3 through Nc ¼ 8. Notice that the requirement of having a minimum of three detections results in a much lower cumulative probability of false alarm ( 12 dB). Target positional uncertainty is minimized by TDS closed-loop tracking so as to result in the beam being pointed at the target under track and the return being approximately centered in the range gate/Doppler filter. An operational loss term for track is included to account for external propagation effects such as multipath fade and atmospheric absorption, as well as residual target uncertainty, if significant. Assuming equivalent transmit and receive antenna areas, these substitutions result in the following sizing relationship to be summed across the Nt TDS tracks: Nt Pa A2 S X ri Ri 4 Loti 4pk N r i¼1 si l2 Ls Ts

(6-9)

where Nt ¼ number of targets under track. As in the search radar range equation, operational losses have been moved to the right side of the equation to accommodate variation in losses due to elevation, scan angle, and other variations. Since an MPARS must multiplex search and track, the left sides of the search and track equations represent performance bounds, with the entire radar resources allocated to search and track, respectively. Unlike mechanically scanned radars, automated beam agility enables an MPARS to dynamically allocate resources among search and track so as to maximize operational utility subject to the sizing metric bounds. Accordingly, the RF aperture should be sized to simultaneously satisfy the search and track sizing requirements metrics such that X S R4max Wi Losi ds PA i 4pk i¼1 N s Ls T s i T fsi ri

(6-10a)

261

262

CHAPTER

Multifunction Phased Array Radar Systems

6

and Nt dt PA2 S X ri Ri 4 Loti 4pk N r i¼1 si l2 Ls Ts

(6-10b)

where ds and dt ¼ the duty cycle allocations to search and track, respectively, such that d ¼ ds þ dt in the absence of other radar functions. In order to exercise this functionality, the MPARS must employ a resource management process as described subsequently that schedules waveforms and beam rates to meet the operational search and track requirements. In addition to meeting the nominal sizing requirements, resource management can enable the radar to shift resources between search and track in response to time-varying loading demands. For example, volumetric search can be slowed in order to divert resources to horizon search or heavy track loading.

6.4

ESA OVERVIEW

6.4.1 Array Principles

q

d inq

Aperture Plane

–1 (N

Plane Wave

)d

sin

q

ds

FIGURE 6-3 ¢ Relationship between Phase Taper Geometry and Beam Steering.

Boresight Direction

The principles of ESA beam scanning are summarized in Figure 6-3 using the onedimensional case of a narrowband N-element array for ease of understanding. An electromagnetic plane wave impinging along the linear array imposes an element-toelement phase differential proportional to the element separation projected along the angle-of-arrival (AOA). The element-to-element phase differential Dj imposed by AOA

6.4

ESA Overview

qAOA for an element spacing d and carrier wavelength l along the plane of the scan is given by d Dj ¼ 2p sin qAOA l

(6-11)

Hence, an impinging plane wave imposes a linear phase taper across the array with a cumulative run-out of (N – 1) Dj rad. A beamformer with fixed-phase weights applied to each element would encounter destructive interference when summing samples of a plane wave received from outside the near-boresight region where Dj ? 0. The narrow angular extent over which constructive interference occurs is the antenna mainlobe, while the region of destructive interference corresponds to the antenna sidelobes. The only way to scan the beam with an antenna employing a fixed-element phasing is to mechanically steer the boresight of the antenna. An ESA scans by commanding element-level phase shifters to the conjugate planewave phase taper for a commanded AOA, the steering angle. The resulting element signals are then summed in a beamformer to produce the antenna output. The ESA effectively serves as a spatial filter matched to an element-sampled plane wave propagating along the designated AOA with a resultant gain proportional to the number of contributing elements N. On transmit, the contributions of each radiating element coherently sum in free space along the AOA corresponding to the commanded beampointing direction. The individual elements possess phase shifters with settings computed and commanded by a beam-steering computer that implements the real-time pointing commands of a central data processor executing the radar control program. Hence, the beam can steer to another position as rapidly as the phase shifters can be reset. Beam-steering times in modern arrays employing PIN diode phase shifters are typically less than 10 ms as determined largely by the period required for beam-steering computations. In a modern AESA, the beam-steering computer typically maintains and implements compensation tables for errors measured during the array calibration process. Hence, the commanded setting of a given phase shifter takes into account the composite measured error of that element path through the array as a function of frequency and array temperature in order to closely approximate the desired element-level taper. The far-field pattern of an aperture antenna possesses a Fourier transform relationship with the excitation function (current distribution) across the aperture [6]. The antenna elements are intended to represent samples of the ideal excitation function subject to hardware errors. The discrete Fourier transform of the element amplitude and phase samples is termed the array factor. If the aperture is undersampled by spacing the elements too far apart relative to the wavelength, grating lobes form analogous to aliased mainlobes. Suppressing all grating lobes within 90o of the mainlobe would mandate that the elements be spaced at l/2 in the designated scan plane. From the Fourier transform relationship, we can infer that amplitude weighting across the aperture can suppress the far-field sidelobes. Since weighting introduces mismatch relative to spatial matched filtering of an impinging plane wave, there is a corresponding S/N loss and mainlobe broadening. Weighting does not significantly suppress grating lobes, as they are aliased replications of the mainlobe. In addition, amplitude and phase errors – deviations from the desired excitation function – produce

263

264

CHAPTER

6

Multifunction Phased Array Radar Systems

spurious-angle sidelobes that are not mitigated by weighting and that impose a corresponding ‘‘floor’’ on achievable sidelobe suppression, as addressed in the next subsection. Phased array antennas are constructed of directive radiating elements so that the actual far-field antenna pattern is the product of the array factor pattern and the element pattern [7]. Elements are generally designed to possess identical electrical characteristics so that the element pattern may be generally represented as corresponding to that of a single radiator embedded in the array electrical environment as a function of AOA, neglecting slight variations due to manufacturing tolerances as well as proximity to the array edge. The element pattern is fixed in angle and centered on boresight, the direction orthogonal to the plane of the aperture. The element pattern’s extent determines the effective scan extent of the array. The gain of the array at a given AOA is then the corresponding gain of the element pattern multiplied by that of the array factor. Assuming that the radiating element is designed to support a full field of view (FFOV), the maximum element pattern gain can be approximated as 10log(p) 5 dB. The maximum gain on boresight, the direction perpendicular to the plane of the array, for a uniformly excited array of N FFOV elements is then approximately 10log(pN) corresponding to the directivity of the aperture as described later. As the ESA beam is steered off boresight, the projected aperture extent decreases, resulting in beam broadening and loss in gain. The two-way transmit/receive element gain of a conventional array can generally be approximated as rolling off with the cosine cubed of AOA. This approximation includes consideration of the element pattern rolloff as well as the projected antenna aperture area. The element pattern extent defines the field-of-view over which the antenna can electronically steer the beam. The element pattern suppresses grating lobes outside this extent [8]. The polarization of the element pattern determines that of the array antenna for a given AOA. Hence, the polarization of a phased array varies with scan angle according to the element pattern characteristics. Polarization agility or diversity requires a switchable radiator at each element, which tends to motivate fixed polarizations for ESAs to minimize RF aperture costs and losses. Parallel beamformers are typically used to provide monopulse difference channels as well as the sum channel described earlier. The monopulse difference channels, commonly termed the delta channels, enable the position of a resolved target to be estimated to within a fraction of a beamwidth in azimuth and elevation. Neglecting correlated errors, the noise-limited AOA estimation precision standard deviation sAOA can be estimated using typical design parameters in a given dimension by the expression sAOA

bo pffiffiffiffiffiffiffiffiffiffiffi ffi km 2S=N cosðqÞ

(6-12)

where bo ¼ beamwidth defined by the 3-dB roll-off mainlobe resolution on boresight, km ¼ the monopulse slope of the antenna based on its design and calibration (typically 1.2–1.8), and q ¼ commanded scan angle in that dimension.

6.4

ESA Overview

6.4.2 Antenna Impacts on MPARS Performance As indicated in the previous section, the mainlobe extent, bq – the antenna beamwidth – of an ESA is commonly estimated as the on-boresight beamwidth divided by the cosine of the aperture extent projected along the commanded AOA. The boresight beamwidth is broadened from the Rayleigh resolution limit according to a weighting factor aw imposed by a sidelobe weighting function. While this beam broadening can be derived through interaction of the array factor and element pattern, it also can be visualized as a consequence of the foreshortening of the projected antenna extent with increasing scan angle. This relationship can be summarized as bðqÞ ¼

bo aw l ¼ cosðqÞ ðN d cosðqÞÞ

(6-13)

From Equation (6-12), there is a corresponding degradation in angle measurement precision with increasing scan angle. The gain of an FFOV ESA can be calculated on boresight as 1 0 l l 4p Nx Ny B 4pA 2 2 C C ¼ 10 logðN pÞ dB (6-14) G ¼ 10 log ¼ 10 logB 2 2 @ A l l for the case of a uniformly illuminated planar aperture with each orthogonal scan plane composed of Nx and Ny elements uniformly spaced at l/2. The implicit relationships of Equations (6-13) and (6-14) are depicted in Figure 6-4 as a function of the number of elements in an FFOV array. The maximum gain denotes the boresight directivity, so sidelobe weighting, electronic scanning, or aperture losses would reduce this value accordingly. The equivalent beamwidth approximation presupposes the elements are implemented as a symmetric array producing a pencil beam. Heavy sidelobe weighting or electronic scanning would broaden the depicted beamwidth. For context, several radar systems are slotted below the figure corresponding to their number of elements and organized by frequency band. The number of elements indicates the electrical size of an RF aperture such that the physical extent has been normalized to wavelength. As noted earlier, the composite ESA directivity at a specified AOA is the product of the element pattern and array factor pattern. The resultant scan loss in directivity for FFOV arrays is calculated as a two-way value corresponding to the composite transmit/ receive impact on sensitivity, typically using the element factor approximation Ls 30 log½cosðqÞ:

(6-15)

The resultant scan loss is plotted in Figure 6-5 along with the corresponding average loss taken between the designated scan angle and boresight. Since the scan losses are well defined, the radar control program may command longer pulses or additional pulses for beam positions at large scan angles to compensate for this loss in order to maintain requisite sensitivity. The peak loss indicates performance at a given scan angle, while average loss should be used to estimate impact on long-term duty cycle usage to support

265

CHAPTER

Multifunction Phased Array Radar Systems

6

10.0

60

FIGURE 6-4 ¢ Antenna Gain and Equivalent Beamwidth with Exemplar Radar Systems.

Maximum Gain (dB)

50 40 1.0

30 20 Directivity Beamwidth

10 0 0

5,000

VHF

Daryal

UHF

BMEWS

25,000

0.1 30,000

Aegis

S-band C-band

PATRIOT

X-band

Fighter Radars

TPY-2

8 7 Two-Way Scan Loss (dB)

FIGURE 6-5 ¢ Peak and Average Scan Loss Presupposing cos3 Roll-off.

10,000 15,000 20,000 Number of Elements (N )

Equivalent Beamwidth (deg)

266

6

Peak

5 4 3 Average

2 1 0 0

10

20

30 Scan Angle (deg)

40

50

60

uniform search performance across the field-of-view. The scan loss must be included in estimating search and track operational losses as well as high-fidelity, beam-by-beam simulations. Distribution of the RF beamforming and amplification process offers graceful degradation of radar sensitivity as a function of component failure. A single-channel module failing on transmit would cost the RF aperture its contribution to transmit power generation and transmit gain, while a module failing on receive would detract from the

6.4

ESA Overview

267

effective receive antenna aperture area. Thus, the S/N loss in dB from Nfail element-level failures of a uniformly illuminated array is given by N Nfail Lfail ¼ 10i log ¼ 10i logð1 Pfail Þ (6-16) N where Pfail ¼ corresponding probability of an element-level failure, i ¼ 1 corresponds to receive-only T/RM failure, i ¼ 2 corresponds to transmit-only T/RM failure, and i ¼ 3 corresponds to joint transmit and receive T/RM failure. This relation presupposes a uniformly illuminated array, so it should be modified to accommodate aperture weightings. Outage loss should be computed according to anticipated failure statistics and included in the nominal system loss factor. As plotted in Figure 6-6, assuming a worst-case failure mechanism in an AESA where a given T/RM fails on both transmit and receive and uniform illumination distribution, some 20 percent of the modules would have to fail before radar sensitivity is reduced by ~3 dB. In addition to this inherent graceful degradation due to the effective parallelism of transmit and receive chains, the attendant long life-times of solid-state components compared to tubes as well as the superior reliability of solid-state low-voltage supplies over the high-voltage power supplies required for transmitter tubes provide high mean times to failure at the component level. As previously noted, the excitation current distribution on the antenna aperture and the resulting far-field pattern are related by the spatial Fourier transform so that aperture tapering can be used to suppress sidelobes. More generally, amplitude/phase illumination patterns can be computed to synthesize arbitrary antenna pattern characteristics while simultaneously steering the mainlobe. In addition, deviation from the commanded illumination pattern due to amplitude and phase errors or outages imposes a corresponding error sidelobe floor that cannot be reduced with array tapering. Wang provides a useful approximation formula for estimating the sidelobe error floor as a function of residual element-level error and probability of randomly distributed failures, as well as 8

FIGURE 6-6 ¢ Outage Losses Due to Failures within Uniformly Illuminated AESA.

Transmit and Receive Failure

7

Sensitivity Loss (dB)

6 5 4 Transmit Failure 3 2 Receive Failure 1 0 0.0

0.1

0.1

0.2

0.2

0.3

0.3

Portion of T/RMs Failed

0.4

0.4

0.5

0.5

268

CHAPTER

6

Multifunction Phased Array Radar Systems

addressing a number of systematic aperture error effects on sidelobe performance [9]. His expression for random error includes mechanical errors in element placement as well as amplitude and phase error due to residual calibration errors. Adopting Wang’s notation for consistency and employing the assumption of small residual errors results in characterizing sidelobe error in dBi as 2 3 2 2 2 2 Pfail þ sa þ sq þ k sm Þ7 6 SLLi 10 log p Pfail þ s2a þ s2q þ k 2 s2m 10 log41 Pfail þ 5 2 p (6-17) where k ¼ 2pd/l, sa ¼ standard deviation of element-level amplitude error, sq ¼ standard deviation of element-level phase error, and sm ¼ root sum square of standard deviation of random element placement error. By way of an example, in order to attain an error sidelobe floor of –15 dBi in the absence of any mechanical error or element failures, it would be necessary to jointly hold amplitude and phase error to 0.1 dB and 0.1 rad, respectively, in the absence of any mechanical fabrication error. A 2 percent failure rate would raise the error floor to ~ –10 dBi. The amplitude, phase, and mechanical errors represent residual values after calibration and compensation techniques have been utilized to eliminate systemic bias-like errors. The associated error budget depends on the accuracy of antenna characterization, the stability of the errors over time, temperature, frequency, and the fidelity of the compensation process implemented by the beam-steering subsystem. The beam-steering fidelity is largely determined by the number of bits available for amplitude and phase element-level control as well as the adequacy of array characterization data to address the span of array operational conditions. An AESA aperture will likely be constructed from line-replaceable units that drive adjoining blocks of elements so that failures may be highly correlated rather than randomly distributed, as presupposed in Wang’s expression. Such correlated failures tend to manifest as increases in the sidelobes near in to the mainlobe, and are of particular concern in rejecting clutter and external interference. In principle, the element-level control of an AESA enables arbitrary pattern synthesis, but in practice, this capability must be supported by alignment and calibration techniques that maintain performance after deployment.

6.5 RADAR CONTROL AND RESOURCE MANAGEMENT Resource management is implemented via a radar control program that plans and schedules the MPARS operations. Planning consists of formulating a course of action to achieve specified objectives while meeting operational and technical constraints. Mission requirements and resource constraints generally dictate cascading levels of action

6.5

Radar Control and Resource Management

decisions with interactions among them. In some implementations, radar action planning is conducted at the weapon system or platform level as a component of mission-level processing tasks. However, implementing an organic planner within the MPARS software is desirable to accommodate long-term or complex operations such as integrated electronic and mechanical beam steering, interleaving resource-intensive data collection to support target identification or electronic protection functions, and interceptor support during an engagement. The planner provides a coarse sequence of radar operations to be executed over a specified time period that is assessed to be within the radar resource capacity. Scheduling is essentially the assignment of radar resources to implement the planner tasks. The scheduler operates according to optimization criteria and constraints dictated by platform and human safety, mission requirements, and radar technical characteristics. Priority-based approaches are used in modern MPARSs, in contrast to some legacy systems, which sometimes used a round-robin-type scheduler that essentially assigned search and track actions to fixed time slots. The scheduler typically provides a precise sequence of action commands for the radar hardware and application software to implement over an action interval or resource period that is significantly shorter than the planning interval. There should also be a means of the scheduler providing feedback to the planner when tasks must be deferred due to resource constraints or preemption. Multiple algorithmic means for implementing planning and scheduling have been investigated that include template techniques, dynamic programming, auctioning, and neural networks. The computational complexity of planning and scheduling combined with the need to support real-time radar operations generally dictate a suboptimal approach that incorporates heuristic elements. Extensive software testing must be conducted under a span of target loading and environmental conditions to establish stability and reliability in a real-time processing environment. The radar control program is also responsible for diverse functions, including health and status monitoring of the radar and coordination with higher-echelon entities, such as weapon system control programs and netcentric operations. While the interaction of a human operator can potentially enhance operational configuration management, countermeasure mitigation, and performance monitoring, the composite decision rate imposed by transmitting and receiving hundreds to thousands of dwells per second while maintaining track on tens to hundreds of targets grossly exceeds the human capacity to participate directly in the control process. The radar control program must define waveform characteristics, carrier frequency, beam-pointing coordinates, time slot allocation, and signal processing parameters for each dwell.

6.5.1 Radar Resources and Constraints The principal radar resources managed by the radar control program are ●

timeline occupancy,

long-term duty cycle, and potentially

signal/data processing capacity.

Timeline occupancy refers to the portion of the radar operating timeline that is occupied by transmit or receive actions and attendant supporting functions such as calibration pulse injection. The occupancy O for a transmit and receive dwell pair using

269

270 FIGURE 6-7 ¢ Radar Transmit and Receive Occupancy.

CHAPTER

6

Multifunction Phased Array Radar Systems

Transmit Timeline

Receive Timeline

Pulse Width

Range Window

a pulsed waveform of duration t, range window DR, and pulse repetition interval T, assuming sequential dwells, can be defined as DR DR ðtÞtransmit þ t þ 2 c receive 2 t þ 2 c 2DR ¼ O¼ ¼ 2d þ (6-18) T T Tc As indicated in Figure 6-7, the radar must operate for an interval corresponding to the sum of the transmitted pulse duration, received pulse duration, and range window extent over which target detections are accepted for each dwell. The achievable occupancy with a heterogeneous mixture of waveforms and functions is generally less than unity, as there is ‘‘dead time’’ imposed by the inability to avoid timing conflicts among transmit and receive actions. The inverse of the search beam rate imposes an upper bound on sequential transmit/receive dwell duration. If the search rate must increase corresponding to enlarging angular extent of surveillance or responding to an operational need to reduce search frame time, the resource manager may be unable to support the pulse duration required for the specified sensitivity or the full range window extent. Interleaved transmit and receive dwells can partially mitigate this constraint. Dedicated pulse Doppler radar operation may result in near-unity occupancy since the transmitter and receiver operation are continually alternated within a coherentprocessing interval (CPI) over each dwell. The dwells may be nearly contiguous, with fixed repetitive time slots assigned to specific search and track operations. However, the composite dwell rate and operational flexibility of such dedicated pulse Doppler phased array radars is generally less than an MPARS employing a diverse waveform repertoire. MPARSs may be constrained by timeline occupancy in practice. Their inherent flexibility and adaptability enable the designer to support multiple users and functions, but as loading increases, time-critical tasks can no longer be executed as promptly as desired. In particular, conducting volumetric surveillance or horizon search at frequent update intervals imposes onerous timeline occupancy demands. Specialized measurement dwells to support semi-active guidance, extended detection range, electronic protection, or noncooperative target identification are used infrequently but tend to dominate short-term radar resource usage when they are utilized. Long-term duty cycle denotes the portion of time over which the transmitter operates as computed across many dwells. Long-term duty cycle is ultimately constrained by the RF aperture cooling capability, as well as prime supply limits. The radar transmitter design may determine the short-term duty cycle by constraining maximum pulse duration due to associated energy storage demands required to maintain the transmitted pulse shape. Signal/data processing resources include computational rate, memory storage, and data transfer among subsystems. Signal processing generally denotes high-throughput/ low-latency functions such as pulse compression, Doppler filtering, and detection

6.5

Radar Control and Resource Management

processing that typically operate dwell to dwell. Data processing generally denotes software processes that operate over a relatively long duration and over many dwells, including radar control, tracking, and user support functions. Depending on the radar architecture, signal and data processing functions may be integrated in a common software/hardware environment or assigned to separate processing subsystems. Designers generally intend MPARS to operate virtually unconstrained from processing limits. However, growth in mission requirements and evolving software complexities often force processing performance limits over the life cycle of the radar. Since search and track performance metrics are proportional to average power radiated, the radar scheduler attempts to attain the full long-term duty cycle to maximize radar performance, but may be constrained by timeline occupancy. Occupancy constraints are particularly severe for AESA architectures that often employ large long-term duty cycles, say 20 percent to 30 percent, to compensate for the relatively low peak power of solid-state high-power amplifier (HPAs). As implied in Figure 6-7, the maximum range window extent must decrease as the pulse duration increases for a fixed dwell repetition interval. However, operational requirements may mandate a minimum acceptable range window and maximum dwell repetition interval in order to sweep a specified search volume in the requisite period to prevent an unobserved target transit. Increasing the aperture size to improve search and track performance increases search occupancy demands, as more beams are required to cover the requisite search volume as the beamwidth decreases. Occupancy constraints may be mitigated by techniques that include: ●

parsing search and surveillance functions up to assure that time-critical search is implemented only in specified regions using narrow fence rasters, defocusing or rapidly re-steering the transmit beam to cover multiple beam positions while maintaining receive coverage by generating multiple simultaneous beams using digital beamforming (DBF) or a specialized analog beamformer, and dividing surveillance among multiple radars or other sensors so as to off-load search demands.

This resource management process must be conducted under operational constraints as well as hardware limitations. The transmit duty cycle may be sharply constrained in specified angular sectors due to RF exposure safety limits. Transmission may also be inhibited in order to avoid own-platform detection and targeting by adversary passive sensors. The transmitter design may constrain maximum pulse duration so that the scheduler must compensate by utilizing burst waveforms in order to achieve the requisite sensitivity in a given beam position. Beam-steering computation and subsystem/ control processing may also impose latency that detracts from the achievable occupancy.

6.5.2 Resource Management Implementation The purpose of the automated radar control process is to efficiently allocate resources to fulfill specified mission requirements while conforming to operational and technical constraints. Modern MPARSs generally employ some manner of priority-based scheduling process that adaptively allocates radar resources to competing demands, such as supporting search and track operations. Figure 6-8 provides a means of visualizing this control process as a cyclic decision and execution flow. A radar action corresponds to

271

272 FIGURE 6-8 ¢ Resource Management Process.

CHAPTER

6

Multifunction Phased Array Radar Systems

Search Verify / TI Track / Illuminate Guidance Maintenance Radar Action Request Queues

Radar Executive • Priority Control • Queue Polling • Event Update • Resource Constraints • Operations Data • Radar Scheduling • External Interfaces Generate Radar Commands Radar/Computer Interfaces Radar • Timing Action • Control Signals • Data Formatting Detection & Estimation Processing Track Association Processing

Operator Interaction C^3 Interfaces

Track File Ensemble Environmental Map

Data Update Track Update

Data Evaluation and Radar Request Generation

the transmission or reception of a given dwell so it involves the highly synchronized operation of the transmitter, receiver, exciter, and digital signal acquisition subsystems. The rest of the radar control and processing process is generally conducted within a general-purpose processor hosting the radar control program and supporting application software such as track filtering and performance/environment monitoring. The radar control program polls an ensemble of action request queues corresponding to multiple search plan execution, verification, and track initiation (TI) requests; track updates; and weapon guidance uplinks, as well as various housekeeping functions supporting test and calibration as indicated in Figure 6-8. As described earlier, the radar control program iteratively schedules the radar actions to be performed over a specific time slot. The request queues are updated during each interval to reflect search and track processing updates. The priority of any given request can be dependent on a number of factors, but some general considerations follow. Weapon guidance and functions critical to ownship survival/safety are given the highest priority. Weapon guidance includes illumination dwells to support semi-active homing, as well as uplink/downlink operations to support interceptor tracking and midcourse guidance updates. Ownship survival/safety tasks may include navigation functions such as support of low-altitude terrain following onboard an aircraft. As noted previously, MPARSs can be designed to immediately verify the presence of a target with a confirmation dwell injected into the estimated position of target detection. Radar search sensitivity can be improved 1–2 dB under typical conditions by lowering the initial detection threshold and depending on the verification dwell to achieve adequate false-alarm suppression. This two-stage sequential detection process is also termed alert-confirm processing. Upon the target returns successfully passing both the search and verify dwell detection thresholds, the radar control program schedules a track initiation (TI) sequence. Typically, this is a high dwell rate of limited duration intended to produce a stable track filter initiation and provide high-confidence measurement-to-track assignment during the initial few updates when the track is relatively coarse.

6.5

Radar Control and Resource Management

Track firmness mandates that the radar tracker accurately predict the position of the target to well within a beamwidth at the time of the next track update. This prediction requirement mandates the need for target velocity estimation. The estimated velocity error standard deviation is inversely proportional to the square root of the time in track so that initial track accuracy is relatively poor in motivating the TI sequence. Verification and track initiation dwell sequences typically represent the second tier of priority since maximizing track initiation range is a critical performance parameter to the operational community. Upon completion of the track initiation sequence, the target track is maintained by periodic track updating at a lower rate typically of no more than a few measurements per second. Position and velocity estimation error generally decreases over the track until attaining a level determined by measurement quality, track filter characteristics, and target maneuvering. Track updates may be assigned a priority depending on identification, maneuver behavior, engagement status, and range. In environments featuring heavy aircraft traffic, only targets of a specified level of priority may be assigned dedicated TDS track beams. In order to conserve radar resources, low-priority targets may be coarsely tracked using TWS techniques collected from one search frame to the next. Search tends to be given less priority than TDS track, as any given search dwell is not time critical in contrast to the process of maintaining track with requisite confidence and accuracy. Unlike mechanically scanned radars with fixed transmit/receive beam patterns, the search pattern of an MPARS can be divided into multiple segments with independent waveforms and revisit rates to enhance performance to provide some degree of adaptivity to threat characteristics and local environmental conditions. As noted, an MPARS may employ a relatively slow volumetric search requiring hundreds of beam positions to cover the range and angular sector, potentially containing targets over a period of tens of seconds. In contrast, the horizon region, where targets enter the radar field-of-view, will likely be searched over briefer intervals on the order of seconds. If the interval required to search through a designated volume threatens to exceed the minimum estimated period required by a target to transit that region, the radar control program may be designed to increase search priority at the cost of decreasing target update rates or even dropping low-priority targets from TDS tracking. The radar may be designed to maintain maps of interference sources, such as clutter compiled from processing search returns. These interference maps may support waveform selection and signal processing control. For example, pulse Doppler waveforms would be used where required for clutter rejection. Adaptive resource management encourages usage of sophisticated track filtering. Simple fixed-coefficient polynomial smoothing filters have been successfully used in legacy MPARSs. However, alpha-beta filtering and other simple polynomial smoothers mandate fixed update intervals and do not provide indications of track quality. From the resource management perspective, one would prefer to be able to advance or retard the track update interval in order to accommodate the scheduling process and achieve higher occupancy. In addition, a real-time estimate of track quality can be useful in the prioritization process of the radar control program. Updates on firm accurate tracks can be deferred, while the update interval should be decreased on unstable tracks or in the event of a missed update detection. Kalman filtering techniques are often used in MPARS tracking. This technique accommodates variable update intervals. The associated covariance matrix also provides a means of estimating track quality, as the diagonal axis terms denote the track filter’s

273

274 FIGURE 6-9 ¢ Resource Management Distributed over Action Intervals.

CHAPTER

Multifunction Phased Array Radar Systems

6

Planning Interval Control Loop Closure

Quick Reaction Nominal

Scheduling Interval

Load Radar Command Buffer (Schedule)

Execute Trasmitter Receive Action T

R

T

T R R

Time Signal Data Processing (Including Plan)

“Quick Reaction” Task Nominal Task Cycle

estimate of the state vector variance. Based on the current covariance matrix and estimated measurement variance, the difference between the predicted target position and the update measurement assigned to that track is adaptively weighted to update the tracker state estimate. Maneuvering characteristics that cannot readily be represented in the track filter typically limit the degree of position and velocity smoothing attainable. Since Kalman filtering performance is susceptible to measurement bias errors and target state model misrepresentations, the degree of smoothing is deliberately constrained by introducing plant noise into the update process. Track filters designed for tracking nonmaneuvering targets can reduce the variance of the radar position measurement by over an order of magnitude over a long-duration track. Improving performance against maneuvering targets is motivating more robust track filtering techniques that may fuse multiple models of target motion in the track update process. The control and execution process is depicted from a timeline perspective in Figure 6-9. The radar timeline has been divided into arbitrary slots to facilitate the common use of non-real-time operating systems to support the radar control program. On the top line, transmit and receive dwells are scheduled for the next time slot and then used to command the radar hardware on the second level, which corresponds to real-time radar hardware operation. The third level corresponds to signal and data processing that result in updated radar action requests for subsequent scheduling. The artifice of scheduling radar actions over finite time slots reduces the computational requirements that would be imposed by optimally scheduling radar actions from dwell to dwell. The scheduler efficiency generally increases and the requisite computation demands decrease with increasing time slot duration. However, the latency of the radar – the time required for it to react to new target detection and other significant events – increases with increasing time slot duration. The scheduling processing efficiency can potentially be improved within these constraints by utilizing scheduling intervals of variable duration. The scheduler may be designed to support ‘‘quick reaction’’ tasks where high-priority dwell requests, such as verifies, are inserted into the next scheduling interval as depicted earlier.

6.5

Radar Control and Resource Management

An MPARS typically implements a repertoire of waveforms matched to specific functional and energy needs. Waveform characteristics consist of modulation format, bandwidth, and duration, as well as pulse repetition interval for burst waveforms. The waveform bandwidth determines the range resolution, while the duration determines signal-to-noise ratio for a given target RCS and range. Burst duration determines Doppler resolution, as it defines the coherent-processing interval. In general, the radar control process chooses the waveform characteristics to minimize dwell time on target so as to enable resource availability for other tasks. For example, a surface-based radar system may restrict use of pulse Doppler waveforms to low-elevation search and track operations where clutter rejection requirements mandate this capability. Single-pulse or moving target indicator (MTI) waveforms would be used at higher elevation angles where the transmit/receive sidelobe suppression is adequate to attenuate surface clutter or the surveillance space is beyond the nominal clutter horizon. While MPARS development emphasizes the RF aperture and the radar control and application software, the need for dwell-to-dwell agility in waveform, frequency, and signal processing parameters also significantly impacts subsystem design. The exciter and waveform generator must be capable of changing pulse duration, bandwidth, and carrier frequency within the beam-steering interval without attendant loss in stability to avoid generating spurious signal components. The signal processor must be capable of efficiently accommodating a heterogeneous series of computations, as waveforms vary from dwell to dwell. Finally, the timing and control interface must be capable of distributing the subsystem commands and maintaining strict synchronization throughout the radar system under all possible dwell sequences.

6.5.3 Multiple Target Tracking Considerations in Radar Control Legacy MPARS may track multiple targets by essentially decomposing the high-priority target set into an ensemble of individual noninteracting track tasks subject to the resource management process. Under this approach, each target under dedicated track is serviced with an injected TDS transmit/receive dwell at a fixed update interval. However, this approach can lead to both performance degradation and inefficient resource usage in dense target environments or other stressing conditions. MPARS control and operation inherently present multiple target tracking challenges in the context of both measurement resolution and measurement-to-track data association [10]. In order to obtain measurements of suitable quality for supporting track or identification tasks, it is imperative that the targets – and sometimes the composite primary scatterers – must be well resolved from other returns in range, Doppler, or angle. In practice, the range and Doppler resolution capabilities are much finer than that afforded by angular resolution. As a general rule, the practical resolution achieved between targets or scatterers of comparable RCS is about twice that predicted by the Rayleigh limit [11]. Joint resolution in range, Doppler, or angle may be required to ensure that returns from strong target scatterers or clutter do not mask weak target scatterers. Waveform design and hardware performance bound the achievable rejection of returns outside the range/Doppler mainlobe, as commonly described in terms of range sidelobe and Doppler sideband suppression, which delineate rejection of resolved returns. Air and missile defense often imposes scenarios where targets are poorly resolved from one another or from clutter for significant durations. In order to preserve radar tracking performance in these circumstances, it may be necessary to utilize waveforms

275

276

CHAPTER

6

Multifunction Phased Array Radar Systems

with superior range (wider bandwidth) or Doppler (longer duration) resolution. Even in the presence of adequately resolved target returns, measurement-to-track data assignment may challenge the radar processing. Mechanically scanned single-target tracking radars effectively ‘‘stare’’ at targets with a resultant high data rate. MPARSs update tracks at lower rates – but still much higher rates than rotating surveillance radars – so that an update dwell could produce multiple measurements that could potentially be associated with a given track. Robust data assignment techniques have been developed that significantly improve track stability at the cost of additional processing demands. Ideally, the tracking process should be highly integrated with the resource management routines, including track file management subprocesses such as initiation, updating, and termination, as well as collateral functions such as redundant track detection and elimination and target counting functions. For example, the data assignment process should be capable of updating multiple independent tracks with a single transmit/receive dwell if there are multiple detection reports within the mainlobe. This capability can mitigate the impact of clutter and countermeasures, as well as enable tracking in dense target environments. The term track while track (TWT) denotes the capability to collect independent measurements and update multiple independent tracks with a single transmit/receive dwell.

6.6

MPARS TECHNOLOGIES

Modern MPARSs generally employ active electronically scanned arrays (AESAs) due to associated benefits of performance, availability, technology base leverage, and life cycle affordability. As illustrated in Figure 6-10, AESAs employ an array of element-level FIGURE 6-10 ¢ RF Aperture Depicting AESA with Digital Beamforming.

EXCITER/WAVEFORM GENERATOR

Digital Beamforming -------Signal/Data Processing

ANTENNA Power Conversion & Transmit/ Conditioning, Receive Cooling, Modules Beam Steering

RECEIVERS Filtering / Amplification, Down Conversion, Analog-to-Digital Conversion

6.6

MPARS Technologies

solid-state transmit/receive modules (T/RMs). Each radiating element is supported by a dedicated channel, which includes a transmit high-power amplifier (HPA), receive lownoise amplifier (LNA), transmit/receive protection circuitry, and common transmit/ receive components such as a phase shifter and attenuator [12]. In practice, multiple T/RM channels may be integrated into a single assembly to reduce production costs. The T/RMs are supported by an RF aperture infrastructure to provide distributed direct current (DC) power conversion and conditioning, cooling, and beam steering as well as mechanical housing. The cost of this infrastructure is typically on the same order as the total T/RM cost in modern systems [13]. While modern AESAs impose lower life cycle costs than functionally equivalent tube-based mechanically scanned radars, initial affordability is a continuing concern. The peak power output of solid-state HPAs is multiple orders of magnitude below that of tube transmitters. For example, off-the-shelf products currently provide in excess of 50 W at S-band and 10 W at X-band. AESAs compensate for this component peak power disparity as follows: ●

Element-level HPA mechanization provides a free-space power combining gain of ~10 log(N) dB, where N is the number of AESA elements; HPAs and LNAs directly feed the associated radiating element, with a circulator/ receiver–protector typically being the only significant intervening component, so that the losses associated with an ESA passive beamformer are eliminated; and Solid-state HPAs can generally be operated at substantially higher duty cycles than high-power tubes so that extended-duration waveforms – longer pulses or pulse integration – can be used to maintain single-measurement sensitivity.

AESAs provide improved reliability and availability over mechanically scanned radars, readily increasing mean time between critical failures by over an order of magnitude. The dominant failure mechanisms for legacy radars are commonly transmit tubes, associated high-voltage power supplies, and antenna positioning servos. All three of these mechanisms are eliminated by a fixed-mount AESA, which uses distributed low-voltage power supplies to support its long-life solid-state amplifiers. Moreover, AESA architectures generally minimize single-point critical failure opportunities since performance is relatively insensitive to the loss of individual transmit/receive modules or power supplies, as delineated later. A passive ESA is composed of an array of element-level phase shifters without active amplification on transmit or receive. A passive ESA beamformer is fed by a tubebased centralized transmitter and receiver, so it offers the advantages of beam agility but without significant improvements in reliability over mechanically scanned systems. However, a number of currently fielded MPARSs employ passive ESAs typically using space-fed phased array antennas to reduce beamformer cost over that of corporate feed designs. Modern AESA radars provide superior sensitivity (range) over passive ESA radars for a given aperture area, total system weight, and prime power consumption. Both acquisition and life cycle costs of a modern AESA for given level of system performance have generally declined below that required when using a passive ESA of equivalent system performance. AESA characteristics also lend themselves to flexible form-fit-factor – supporting specialized implementations that mandate low radar cross section, light weight, platform compatibility, and conformal aperture applications. In many mobile applications, installing fixed-mount AESAs is a less complex and costly

277

278

CHAPTER

6

Multifunction Phased Array Radar Systems

undertaking than attempting to mount a mechanically scanning antenna. These considerations are illustrated by airborne early warning radar providers that mate low-cost AESAs with user legacy airframes. The progress in AESA cost reduction can be assessed by the affordability of multiple-function fighter radars being marketed for both new and retrofit aircraft procurements, offering significant performance enhancement at comparable acquisition cost to legacy mechanically scanned radars. Development of air, artillery, and tactical missile self-defense radars employing low-cost AESA technology also represents a major application opportunity exploiting TDS capabilities to multiplex search and track. The trend toward reduced ownership cost may enable MPARS technology to be applied to civilian applications such as integrated air traffic control and weather sensing, where they can offer enhanced performance and reliability. At present, bipolar silicon (Si) technology tends to dominate low-frequency AESA applications at L-band and below, while gallium arsenide (GaAs) component technology dominates microwave AESA applications at S, C, and X. Gallium nitride (GaN) technology appears very promising to provide increased power, bandwidth, and efficiency subject to establishing an adequate production base for integrated devices to achieve affordability. Current and future AESA component affordability depends heavily on leveraging commercial foundry operations, as radar applications are dwarfed by commercial communication markets for solid-state RF technology [14]. The key components of a T/RM are depicted in Figure 6-11. Amplitude and phase control are typically combined in a common transmit/receive circuit, while the HPA and the LNA are dedicated to the transmit and receive paths, respectively. Power conversion/conditioning and beam-steering mechanization are typically distributed across multiple elements. In contrast to a passive ESA, which performs element-level phase shifting but does not support transmit or receive element-level amplification, the AESA elements are fed directly by transmit and receive amplifiers, minimizing associated losses and system noise contribution. T/RMs may be integrated into a single multiplechannel assembly feeding two or more individual radiating elements in order to reduce production cost. Taking into account respective losses, cooling requirements, and prime power conversion efficiencies, a modern AESA provides superior sensitivity to an ESA system for a given aperture area, weight allocation, and prime power input for less life cycle

FIGURE 6-11 ¢ Elements of a Transmit/Receive Module.

HighPower Amplifier Transmit Beamformer

T/R Switch

Phase Shifter

Attenuation

Radiator

T/R Switch

Receiver Beamformer T/R Protection

Timing and Control Logic/Buffer

Power Conditioning

LowNoise Amplifier

6.6

MPARS Technologies

cost. The AESA may subsume on the order of three-fourths of the radar recurring cost in a modern MPARS, so this motivates technology to reduce RF aperture costs through increasing component integration and other means. There are extensive efforts to develop ‘‘radar-on-a-chip’’ technologies integrating transmit/receive RF and digital functions. However, this technology is currently limited to low-power applications [15]. Designing the RF aperture to resist damage-induced failure is also a major imperative, as the high-cost AESA components are relatively exposed in the RF aperture with limited protection against physical or electromagnetic damage. AESA designs facilitate digital beamforming (DBF) since received signals are amplified at the element level. Modern DBF implementations can aggressively leverage commercial technology performance and affordability while enhancing MPARS performance and improving resilience. DBF effectively distributes the receiver subsystem across the array, with individual elements or small groups of elements, termed subarrays, feeding individual receive chains as depicted in Figure 6-10 [16]. The composite antenna pattern is formed digitally within the signal processing subsystem. Unlike a conventional analog beamformer, digital processing enables multiple beams to be generated by iteratively summing a given received signal set while preserving the full gain of the array for each beam. DBF offers significant operational utility from its inherent capability to flexibly generate multiple simultaneous receive beams with arbitrary weighting schemes. Timeline occupancy demands can be potentially mitigated by transmitting over a small angular sector that can be covered by a cluster of receive beams to collect returns in parallel. Adaptive digital beamforming (ADBF) incorporates adaptive interference suppression by integrating sidelobe cancellation and sidelobe blanking while providing a limited mainlobe cancellation capability. DBF is critical in airborne radar applications that must meet stressing subclutter visibility requirements using space–time adaptive processing. The inherent parallelism of DBF mitigates single-point failure mechanisms within the RF aperture. Integrated AESA/DBF apertures enable modular architectures that can potentially be scaled for multiple applications while offering the economy-of-scale benefits of a common technology base. AESA RF apertures generally dominate the recurring cost of a modern radar system, but software development, updating, and sustainment are major concerns in life cycle cost. While previous military software initiatives have provided mixed results, the present trend toward increased use of commercial digital hardware and software development tools and practices appears promising. Emerging open system architecture standards for software should enable graceful rehosting across different generations of hardware and even different computer architectures. In addition, the use of modular and loosely coupled programming practices promises to ease software maintenance and potentially enable code reuse. Extending these techniques to support real-time operation remains a major challenge for MPARS applications. The high development and acquisition costs of high-performance sensors encourage spiral development practices, where incremental capabilities are implemented over the lifetime of the system as adversary capabilities evolve and enabling technologies mature. In the specific instance of AESA radar systems, the high cost of the RF aperture dictates that it will not be modified over the life cycle of the radar system, other than occasional refurbishment of line-replaceable units containing failed components. However, multiple generations of software-based capability enhancement can be anticipated along with supporting processor upgrades. DBF can extend spiral development practices

279

280

CHAPTER

6

Multifunction Phased Array Radar Systems

into the beamforming regime by enabling hardware upgrades to increase throughput and bandwidth as well as algorithmic enhancements. In the face of this trend toward using AESA technology for radar applications, the historical Russian success in low-cost space-fed passive arrays employing centralized transmitter/receiver chains is noteworthy [17]. Over the course of the Cold War, the Soviet Union deployed one to two orders of magnitude more phased array radars than did the United States despite the military technology leadership of the latter. While this comparative emphasis on ESA radars was partially due to asymmetric air defense strategies, Russian industry developed antenna architectures and components designed to minimize acquisition cost while delivering acceptable system-level performance. In addition, they constructed integrated air defense systems composed of multiple ESA radar designs optimized for search, battle management, and fire control, as opposed to the single multiple-function radar approach exemplified by the U.S. PATRIOT system.

6.7

MPARS TESTING AND EVALUATION

MPARS development and testing are far more complex than that for legacy singlefunction radar systems. This complexity is imposed by the interleaving of multiple search and track task sequences, as well as the challenges of fully exercising the full performance range of the ESA and supporting subsystems. A mechanically scanned target-tracking radar can be well tested by a series of tracks conducted against a span of representative targets. In contrast, the tracking performance of an MPARS must be evaluated under realistic resource management conditions that include interleaving of multiple target tracks and search operations and associated verification and track initiation. Data association may be more important in determining track performance than in dedicated mechanically scanned radar systems since MPARSs tend to employ significantly lower track update rates than ‘‘staring’’ sensors, which can only track a single target. Graceful interaction between dedicated TDS tracks and coarse TWS tracks generated from surveillance measurements is also desirable to support load shedding as well as suppress misassociation and wasted verification/track-initiation dwells on targets already under track. In addition to these considerations, testing of MPARS target-tracking capabilities must address the interaction of waveform selection and electronic scanning with measurement quality. Some manner of simulation-based testing is necessary in order to establish MPARS performance. At a minimum, the radar control program and application software should be tested on representative data processing hardware to verify and validate performance. It may also be highly desirable to employ some manner of hardware-in-the-loop implementation that drives the digital processing software/hardware in a real-time environment while emulating radar subsystem interfaces and simulating target/environment returns. As introduced in Section 6.3, high-fidelity simulation efforts are challenged to capture the full span of operational losses due to target positional uncertainty and environmental effects. Simulations of lesser fidelity may entirely neglect the radar measurement modeling process so as to not capture the performance impacts of merged measurements and measure-to-track data misassociation. MPARS simulation accreditation efforts should strive to ensure that radar performance is not represented as excessively optimistic through a combination of targeted analyses and test data anchoring.

6.8

Netcentric MPARS Applications

281

At a major increment in cost, the MPARS hardware-in-the-loop configuration may also incorporate a full-size or subscale RF aperture to conduct testing in which the RF aperture participation is critical. At a minimum, incorporation of a subscale aperture should enable the exercise of the beam-steering system and, if present, DBF subsystem. Motivating imperatives for including RF aperture and the attendant transmission and reception operations include determining performance in clutter-limited conditions and investigating susceptibility to electromagnetic interference and electronic attack.

6.8

NETCENTRIC MPARS APPLICATIONS

Historically, the cost of developing and deploying MPARSs has been motivated by their military utility in the context of a specific weapon system or platform. AMD systems such as PATRIOT and Aegis illustrate this legacy. The MPARS enhances and accelerates the kill chain sequence of detection, track, weapon guidance, and kill assessment against a given target, as well as supporting simultaneous engagements against multiple targets. MPARS cost was somewhat mitigated by subsuming the roles of multiple mechanically scanned radars that would otherwise be required. Future MPARS development and deployment costs will be largely motivated by netcentric applications, particularly for high-cost implementations. Netcentricity denotes the exploitation of a robust, globally interconnected network environment in which data is shared seamlessly among users, applications, and platforms [18]. MPARS military utility will increasingly be evaluated in terms of the capabilities that it brings across multiple platforms and in a system-of-systems context, as exemplified in Figure 6-12. Netcentric data sharing is performed among application domains as well as within a given domain, as denoted by the netcentric data products depicted as examples. The availability of high-reliability, low-latency, broadband data-links among sensors and users is a necessary but not sufficient condition for netcentric operations. The MPARS can potentially provide multiple data products that are all referenced to a distributed tracking network that assimilates inputs from multiple sensors to generate a

Air Warfare • Air-to-Air • Air-to-Surface

Targeting

SIAP

Targeting

Air & Missile Defense • Acquisition • Tracking • Characterization

Counter Fire • Search • Track

RSO Catalog

Foreign Launches Threat Tracks

Increasing MPARS Size

RSO Characterization

Space Situational Awareness • Resident Space Object (RSO) • Tracking & Characterization • Launch & Reentry • RSO Catalog Maintenance

BM Early Warning • Surveillance • Attack Assessment

RSO Observations

FIGURE 6-12 ¢ Interacting MPARS Missions and Data Sharing.

282

CHAPTER

6

Multifunction Phased Array Radar Systems

single integrated air picture (SIAP). Tracks must be assigned global identifiers that are reliably correlated with local tracks generated by participating sensors. MPARSs can contribute significant netcentric military utility in the context of this global track ensemble. Netted data processing from multiple TWS radars can increase composite surveillance performance in terms of coverage and track continuity. However, netcentric MPARSs potentially offer markedly improved capabilities in the context of situational awareness and distributed fire control. MPARSs possess the advantages of relatively high data rate and waveform optimization to assure high signal-to-interference ratio and precision measurements subject to accommodating the TDS tracking load. TDS precision tracking against high-priority or stressing targets enhances both accuracy and firmness over that achievable with TWS surveillance radars. Enhanced tracking combined with high-resolution target identification significantly increases situational awareness. Augmented with the capability to rapidly multiplex and tailor search scans, this multiple-target track/identification capability enables an MPARS to develop highquality situational awareness. Transferring this situational awareness to a distributed user base is the foundation of MPARS netcentric military utility. These fire-control-quality tracks can be accessed by external weapon systems to support netcentric engagement modes. For example, a weapon system can use externally provided fire-control-quality tracks to support commit-on-remote, where an interceptor is launched against an incoming threat before the organic fire-control sensor establishes track. Similarly, engage-on-remote utilizes eternally provided tracks to conduct the entire engagement. The capability to rely on coarser, less costly sensors for surveillance enables MPARS resources to be focused on these high-value tracking tasks. Contributions from other sensors may also mitigate MPARS coverage gaps due to line-of-sight blockage, multipath fades, external interference, or other degradations. MPARS netcentric implementation mandates incremental design requirements that can be set in the context of netcentric precepts as summarized in Table 6-2. A key netcentric operational challenge is tasking and prioritization procedures that enable distributed users to modify the behavior of the MPARS in response to their individual needs. The high data quality afforded by an MPARS can readily lead to it being overtaxed by external requests. TABLE 6-2

¢

Netcentric Data Characteristics for MPARS

Netcentric Data Characteristic Visible

Accessible

Usable

MPARS Design and Operational Impacts Both TDS and TWS track data must be integrated into a global track database. Environmental data, such as interference mapping and detection statistics, should also be visible on the network, as well as radar health and status reporting. In addition to accessing global track data, MPARS should respond to external tasking for both radar actions and data products: ●

Radar actions: track prioritization, and special data collection

Data products: measurement reports, tracks, specialized measurement, and reporting outputs

MPARS data should be synchronized with global timing reference and referenced to an external coordinate system. Data products must conform to global track data structures.

6.10

6.9

Further Reading

REFERENCES

[1] W.J. Fontana and K.H. Krueger, ‘‘AN/SPY-3: The Navy’s Next-Generation Force Protection Radar System,’’ 2003 IEEE International Symposium on Phased Array Systems and Technology, Waltham, MA, 14–17 October 2003. [2] J.K. Green, ‘‘F/A-18E/F Active Electronically Scanned Array (AESA) Radar: Lessons Learned from System Development and Integration,’’ 2007 Military Radar Symposium, 27 June 2007. [3] L. Blake, Radar Range Performance Analysis, Munro Publishing Co., Silver Spring, MD, 1991. [4] R.J. Galejs and C.E. Muche, ‘‘Beam and Filter Straddle Losses in an ESA Search Radar,’’ Lincoln Laboratory Technical Report, 31 March 1992. [5] P.J. Kahrilas, Electronic Scanning Radar Systems (ESRS) Design Handbook, Artech, 1976. [6] N. Fourikis, Advanced Array Systems, Applications, and RF Technologies, Academic Press, 2000. [7] R. Mailloux, Phased Array Handbook, Second Edition, Artech House, 2005. [8] W. Wirth, Radar Techniques Using Array Antennas, IEE, 2001. [9] H. Wang, ‘‘Performance of Phased-Array Antennas with Mechanical Errors,’’ IEEE AES Transactions, April 1992. [10] S. Blackman and R. Popoli, Design and Analysis of Modern Tracking Systems, Artech, 1999. [11] F. Daum, ‘‘A System Approach to Multiple Target Tracking,’’ Multitarget-Multisensor Tracking: Applications and Advances, Y. Bar-Shalom, Artech, 1992 [12] B. Kopp, M. Borkowski, and G. Jerinic, ‘‘Transmit/Receive Modules,’’ IEEE Transactions on Microwave Theory and Techniques, Vol. 50, No. 3, March 2002. [13] Y. Mancuso, P. Gremillet, and P. Lacomme, ‘‘T/R-Modules Technological and Technical Trends for Phased Array Antennas,’’ 2005 European Microwave Conference, Elancourt, France, 4–6 October 2005. [14] B. Kopp, ‘‘S- and X-band Radar Transmit/Receive Module Overview,’’ 2007 IEEE Radar Conference, 17–20 April 2007, pp. 948–953. [15] E. Brookner, ‘‘Phased-Array Radars: Past, Astounding Breakthroughs and Future Trends,’’ Microwave Journal, Vol. 51, No. 1, January 2008. [16] V. Chernyak, I. Immoreev, and B. Vovshin, ‘‘70 Years of Russian Radar Industry,’’ 2004 International Conference on Radar Systems, Philadelphia, 26–29 April 2004. [17] D.K. Barton, ‘‘Design of the S-300P and S-300V Surface-to-Air Missile Systems,’’ Located at www.ausairpower.net/APA-Russian-SAM-Radars-DKB.html#mozTocId832616, March 2009. [18] ‘‘Data Sharing in a Net-Centric Department of Defense,’’ Department of Defense Directive, Number 8320.02, 2 December 2004.

6.10

FURTHER READING

T. Jeffrey, Phased-Array Radar Design: Application of Radar Fundamentals, SciTech Publishing, 2009. S. Sabatini and M. Tarantino, Multifunction Array Radar: System Design and Analysis, Artech, 1994. N. Fourkis, Advanced Array Systems, Applications, and RF Technologies, Academic Press, 2000.

283

CHAPTER

Ballistic Missile Defense Radar

7

Melvin L. Belcher, Jr., Georgia Tech Research Institute, Atlanta, GA, USA

Chapter Outline 7.1 7.2 7.3 7.4 7.5 7.6 7.7

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . BMD Radar System Requirements . . . . . . . . . . . . . . . . Radar Development for Ballistic Missile Defense . . . . . BMD Radar Design . . . . . . . . . . . . . . . . . . . . . . . . . . . BMD Radar Performance Estimation . . . . . . . . . . . . . . References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

7.1

....... ....... ....... ....... ....... ....... .......

........ ........ ........ ........ ........ ........ ........

.... .... .... .... .... .... ....

285 292 298 307 312 321 322

INTRODUCTION

A class of radar systems has been developed and deployed to detect, track, and identify targets that are in excess of an order of magnitude greater in range, smaller in radar cross section, and faster than conventional air-breathing targets. These radars perform interrelated missions consisting of ballistic missile defense (BMD), ballistic missile warning (MW), and space situational awareness (SSA). Full field-of-view (FFOV) electronically scanned arrays (ESAs) are employed to interleave search and track functions within a given mission as well as tailor operations across multiple missions. This class of radar can be defined by large physical size, exceptional sensitivity, and high cost. Mechanically scanned reflector-antenna radars are employed to provide affordable sensitivity for niche missions such as test instrumentation and tracking resident space objects but are fundamentally constrained by lack of organic surveillance and target traffic handling. The principal U.S. radars composing this class are listed in Table 7-1. While this chapter is focused on BMD, the MW and SSA missions are also addressed in the introduction because a given sensor may perform aspects of all three missions. In the United States, the Integrated Tactical Warning and Attack Assessment (ITWAA) system detects and tracks strategic ballistic missile threats to the United States. The current generation of early warning radars (EWRs) supporting the ITWAA mission employs active electronically scanned arrays (AESAs) operating in the ultrahigh frequency (UHF) band. While the United States has deployed geosynchronous satellites to detect missile launches from the short-wave infrared (SWIR) signature of the boost phase, EWRs are critical to verifying and characterizing the attack as well as estimating the impact points of the individual reentry vehicles (RVs). The first generation of missile 285

286

Ballistic Missile Defense Radar

CHAPTER

7

TABLE 7-1

U.S. ESA Radars for BMD, Missile Warning, and Space Situational Awareness

¢

Radar Name and Siting

Description

Missions

Capabilities

Eglin (AN/FPS-85) Radar; Eglin Air Force Base, Florida

UHF FFOV dual transmit–receive aperture ESA

SSA

Perimeter Acquisition Radar Attack Characterization System (PARCS); Cavalier Air Force Station, North Dakota Ballistic Missile Early Warning System (BMEWS): Clear Air Force Station, Alaska; Thule Air Force Base, Greenland; and Royal Air Force Station, Fylingdales, United Kingdom PAVE Phased Array Warning System (PAVE PAWS); Cape Cod Air Force Station, Massachusetts, and Beale Air Force Base, California COBRA DANE; Eareckson Air Force Station on Shemya Island, AK Sea-Based X-Band Radar (SBX): Based on mobile floating platform AN/TPY-2; transportable; currently deployed in Shariki, Japan, and in Israel as forward-based radar; > 10 planned Aegis BMD SPY-1; Aegis cruisers and destroyers; > 30 planned for BMD

UHF FFOV ESA

Missile warning, SSA

Primary near-Earth sensor Metric and narrowband signature collection Metric and narrowband signature collection

UHF FFOV AESA:

Missile warning; BMDS upgraded early warning radars: Fylingdales, Thule, and Clear (planned) SSA

Metric and narrowband signature

UHF FFOV AESA; Two faces per site ~1,800 AESA elements per face

Missile warning; BMDS Upgraded Early Warning Radar: Beale and Clear (planned) SSA

Metric and narrowband signature collection

L-band FFOV ESA

BMDS; SSA

Metric and signature collection

Two faces at Clear and Thule Three faces at Fylingdales ~2,560 AESA elements per face

X-band LFOV AESA BMDS on mechanical mount: ~45,000 AESA elements X-band FFOV AESA BMDS: two roles~25,300 AESA forward-based elements radar THAAD

S-band FFOV ESA four faces per ship

BMDS; naval air defense

Metric and narrowband and wideband signature collection Metric and narrowband and wideband signature collection

Metric and signature collection

7.1

Introduction

warning radars employed fixed-beam search radars and mechanically scanned dish radars and was one of the first billion-dollar development projects undertaken by the U.S. defense establishment. The current generation of AESA EWRs provides superior traffic handling and reliability. BMD radars are integrated into the Ballistic Missile Defense System (BMDS) which includes radars in the major elements of Ground-Based Midcourse Defense (GMD), Aegis BMD (ABMD), and Terminal High Altitude Area Defense (THAAD). The confluence between missile warning and missile defense is obvious as both missions require: 1. detection of incoming ballistic missile threats near maximum line-of-sight range; 2. characterization of ballistic missile complexes into major associated components – booster, reentry vehicle, and associated objects as well as supporting target identification functions; and 3. tracking of individual targets with sufficient accuracy to predict their trajectories even in the event of a raid comprised of multiple ballistic missile launches. BMD enforces a more demanding mission on the supporting radar systems. The fidelity of MW characterization and tracking must only be adequate to provide situational awareness to support political and strategic military responses. BMD radars must precisely identify and track RVs so as to support fire control. BMD fire control requires trajectory prediction of the identified target to support long-range interceptor acquisition, which exerts onerous requirements for sensitivity, resolution, and accuracy. Space situational awareness provides comprehensive knowledge of the space object population, including threats, event prediction, and object status as well as tactical picture generation. This mission requires metric data from tracking to support conjunction analysis among resident space objects (RSOs) so as to provide warning of potential collisions. Characterization data are required to aid in identifying objects and assessing their operational state. A global network of radars forms the backbone of the U.S. Space Surveillance Network (SSN) that provides the requisite metric and signature data for SSA. The SSN radars are augmented by ground-based and spaceborne platforms employing optical sensors. Optical sensors can provide superior range against typical space objects but are constrained by requisite solar illumination conditions that limit their utility against the low-altitude regime where RSOs are often in Earth’s shadow. SSN ground-based telescopes are only able to detect solar-illuminated space objects, with the observation sites in darkness. Ground-based radars generally dominate SSA coverage of RSOs at altitudes below 2,000 km, which is why the coverage is termed the low Earth orbit (LEO) regime. The LEO regime is heavily populated by RSOs, including operational satellites, defunct satellites, miscellaneous launch hardware, and primarily orbital debris. Table 7-2 lists the principal fixed-beam and mechanically scanned radar systems currently supporting SSA. Data from these radars as well as optical sensors are integrated with operational and space environment data to provide status, current and future positions, and characterization of the space object population and associated risks and threats. The current very-high-frequency (VHF) fixed-beam continuous wave (CW) Space Fence was fielded in 1961 and was originally known as the Naval Space Surveillance (NAVSPASUR) before it was transferred to the U.S. Air Force in 2004. There is currently an ambitious effort underway to develop a next-generation Space Fence consisting of a modular S-band beam-agile radar system. The SSN also includes a number of mechanically scanned reflector-antenna radars deployed around the world to

287

288

Ballistic Missile Defense Radar

CHAPTER

7

TABLE 7-2

U.S. Detection Fence and Mechanically Scanned Radars Currently Used for SSA

¢

Description

Air Force Space Surveillance System (AFSSS): three transmit antennas and six receive antennas geographically located along the 33rd parallel of the United States from Georgia to California Globus II: Vardø, Norway

VHF fixed beam Detection pattern using bistatic continuous wave

1961

X-band dish radar

1999

Ascension radar: Ascension Island, South Atlantic Ocean Haystack radar: Westford, Massachusetts Haystack auxiliary radar: Westford, Massachusetts Millstone Hill radar: Westford, Massachusetts Advanced Research Projects Agency (ARPA) Lincoln C-Band Observables Radar (ALCOR): Kwajalein Atoll, Marshall Islands ARPA Long-Range Tracking and Instrumentation Radar (ALTAIR): Kwajalein Atoll, Marshall Islands Target Resolution and Discrimination Experiment (TRADEX): Kwajalein Atoll, Marshall Islands Millimeter wave (MMW) radar: Kwajalein Atoll, Marshall Islands

Capabilities

Year Fielded

Radar Name and Siting

Metrics and narrowband and wideband signature Deep space capability C-band dish radar Metrics and narrowband signature X-band dish radar Metrics and narrowband and wideband signature Deep space capability Ku-band dish Radar Metrics and narrowband and wideband signature L-band dish radar Metrics and narrowband signature Deep space capability C-band dish radar Metrics and narrowband and wideband signature

1971 1963

1993 1957

1970

VHF and UHF dish radar

Metrics and narrowband signature Deep space capability

1970

L- and S-band dish radar

Metrics and narrowband signature

1963

Ka- and W-band dish radar

Metrics and narrowband and wideband signature

1983

provide precision metric tracking to support orbital element estimation and to collect signature measurements to support characterization, identification, and monitoring of individual space objects. As noted in Table 7-1, a number of the large UHF phased array radars are dominant contributors to SSA. The first generation of missile warning (MW) sensors deployed in the 1950s employed a similar architecture of fixed-beam detection radars and a small number of mechanically scanned precision tracker but target traffic handling and associated tracking accuracy requirements drove the replacement of these radars with the multiple-function phased array radars listed in Table 7-1. Mechanically scanned radars are not currently used for BMD or MW due to their limited target traffic handling. BMD

7.1

Introduction

metric (prediction accuracy) and signature (number of independent observations) performance benefits from extended observation intervals at moderate measurement rates to assure track firmness and adequate signature sampling to support discrimination. Multiple-function phased array radars provide this mixture of adaptive traffic handling, variable data rates, and extended observation by tracking objects across their FFOV extent. Mechanically scanned radars are effectively staring sensors that are well suited for collecting continuous high-update rate metric and signature data on a single target or closely spaced target set. The superior sensitivity for a given cost of mechanically scanned radars over phased array radars provides them with enduring mission effectiveness for high-altitude orbit and deep-space SSA applications. The SSA community is migrating toward beam-agile radar systems for LEO surveillance as exemplified by the legacy AN/ FPS-85 which tracks some 95 percent of the current Satellite Catalog. While most of the MW/BMD radars are located at northerly latitudes and boresighted to view near transpolar ballistic missile trajectories, the AN/FPS-85 is located in the southern United States and is southerly boresighted. This arrangement significantly increases the number of potential RSO observations. Concerns over improving SSA timeliness, track quality across the space object population, orbital conjunctional analysis accuracy to predict potential collisions, and detection of small space objects all motivate the SSA community interest in developing and deploying phased array radars such as the next-generation Space Fence. Table 7-3 provides a summary of the mission characteristics of missile warning, BMD, and SSA. In general, SSA sensors must support a larger operational range span to cover the orbital regimes than required for BMDS or MW operations. In the absence of anomalous propagation or major geographical features such as mountains, groundclutter returns typically fall below the radar horizon beyond 50 km or so. MW, BMD, and SSA applications share the common attribute that they are designed to attain noiselimited performance since their minimum range of interest is typically well beyond this point. However, at frequency bands below UHF, electromagnetic interference can become a major sensitivity constraint. All three missions impose similar requirements such as mitigating propagation effects. As the radar line of sight is transiting the entire atmosphere, it is necessary to suppress tropospheric refraction effects at low elevation, including diurnal and seasonal variations. Radars operating at L-band and below tend to be susceptible to ionospheric propagation effects, including amplitude and angle scintillation. A nuclear bust in the atmosphere would create an extended region of dense ionization that would more severely block, attenuate, and degrade radar propagation with severity increasing with decreasing frequency [1]. All three mission sets motivate accurate tracking that must produce precise estimates of target position and velocity in an inertia-based coordinate system. In turn, this requires that the radar systems must be well registered in a geodetic coordinate system and correct propagation-refraction effects so as to prevent bias-like errors from degrading systemlevel performance. This capability is critical to extrapolating ballistic target trajectories as well as performing conjunction analyses of space objects to project potential collisions. While the SSA community must deal with ~20,000 resident space objects, these are distributed throughout the immense spatial volume of Earth’s orbital regime, so the primary tracking challenge is correlation of observations (measurements and tracks) with predicted passes of cataloged RSO entries based on their associated ephemeris data. In contrast, ballistic missiles may impose the challenge of closely spaced objects (CSOs) where the booster, RV, or other objects are in close proximity, challenging resolution and

289

290

Ballistic Missile Defense Radar

CHAPTER

7

TABLE 7-3

Comparison of Mission Requirements for Sensors

¢

Mission Characteristic

Missile Warning

Sensitivity drivers

Radar cross section (dbsm) Range (km)

threat complex > 0

targets < 0

1,000–4,000

500–4,000

Metric accuracy (referenced to external coordinate system)

Launch-point estimation Impact-point prediction

Narrowband Wideband

Target traffic handling

LEO SSA ●

Satellites > 0

orbital debris < 0

Altitude < 2,000 km at low-Earth orbits

Noise-limited detection: Operating range generally beyond maximum ground-clutter range Tropospheric refraction Ionospheric propagation; Nuclear burst effects

threat complex > 0

targets < 0

Environment effects

Nominal performance limit

Signature

BMD

Tropospheric refraction Ionospheric propagation Launch-point estimation Impact-point prediction Track and discrimination Interceptor designation

Target classification

Target classification

None

Target identification Countermeasure and clutter rejection

Hundreds of simultaneous Tens of ballistic ballistic missile threats missiles under engagement

Satellite catalog RSO correlation Conjunction analysis (orbit prediction)

Object classification; Debris size estimation

Range-Doppler imaging

~20,000 objects currently tracked in Earth orbits

measurement-to-track data association. Tracking in CSO environments may motivate high-resolution tracking waveforms to ensure firm tracks on individual objects. Narrowband measurements – meaning that individual scatterers comprising a given target are unresolved – can be used to classify the type of object under track. Observed narrowband radar cross-section (RCS) statistics in conjunction with track characteristics can coarsely classify targets. Narrowband data are used for SSA characterization of RSOs as well. For example, empirical relationships have been developed to estimate space debris size from its RCS. Target identification or discrimination within a class is essential to distinguishing between RVs and decoys. Decoys may be deployed from the same launcher as the RV and designed to emulate RV metric and signature characteristics so as to force the missile defense system to waste interceptors against nonlethal objects. Discrimination may require high-resolution measurements such that a given target is resolved into component scatterers. Wideband measurements provide resolved scatterer measurements across the

7.1

Introduction

target range extent while coherent processing across a train of wideband pulse can be used to generate a two- or three-dimensional image of a target [2]. High-resolution radar imaging enables shape and dimensional estimation of the target as well as providing indication of its rotational motion characteristics. Moreover, resolution of individual scatterers can enhance the fidelity of RCS and polarization feature extraction.

7.1.1 Organization This remainder of this chapter expands the description of BMD radars as this sector is currently the focus of extensive development, deployment, and testing efforts as well as posing a number of technical challenges. A brief overview of ballistic missile characteristics and associated defensive architecture concerns are provided for context. The BMD engagement sequence and kill chain are then delineated with focus on the radar roles. A brief account of U.S., Russian, and other international BMD radar-deployment efforts is then provided because upgrading legacy sensors with new software and technology insertion is critical to modern BMD radar efforts. Radar functional requirements and supporting technology considerations are then addressed. The chapter concludes with an account of selected BMD performance metrics.

7.1.2 Acronyms ABMD AESA BMDS BMEWS CEP CSO ESA FFOV GMD ICBM IR ITWAA LDS LEO LFOV MDA MW RSO RV SSA SWIR THAAD UHF VHF

Aegis ballistic missile defense active electronically scanned array Ballistic Missile Defense System Ballistic Missile Early Warning System circular error probability closely spaced objects electronically scanned array full field of view Ground-Based Midcourse Defense intercontinental ballistic missile infrared Integrated Tactical Warning and Attack Assessment launch-detection satellite low-Earth orbit limited field of view Missile Defense Agency missile warning resident space object reentry vehicle space situational awareness short-wave infrared Terminal High Altitude Area Defense System ultrahigh frequency very-high frequency

291

292 FIGURE 7-1 ¢ Ballistic Missile Flight Regimes.

CHAPTER

Ballistic Missile Defense Radar

7

1,600 km Reentry

Warheads and Decoys

Launch of ICBM

1,000

100

Top of the Atmosphere Boost Phase

15

Clouds

Postboost Midcourse Phase Terminal Phase

7.2

BMD RADAR SYSTEM REQUIREMENTS

7.2.1 Overview of the Ballistic Missile Threat Ballistic missiles of varying degrees of capability are available to any established or emergent power as the critical technologies are widely proliferated [3]. The nominal flight path and associated atmospheric regimes of a long-range ballistic missile are depicted in Figure 7-1. A single- or multiple-stage booster is used to launch the ballistic missile according to its range. Short-range tactical ballistic missiles with ground ranges of less than 300 km typically use an integrated warhead and booster. The World War II V-2 and the variants of the descendant Russian Scud exemplify this single-stage class. Multiple-stage, long-range ballistic missiles package the weapon payload into a reentry vehicle designed to protect it as it transits the atmosphere. An accurate RV generally requires an exoatmospheric deployment orientation followed by spin stabilization to minimize atmospheric drag effects upon reentry. Associated debris may be produced from separating launch hardware, expended solid fuel, and other sources that manifest as volumetric clutter. The composite ensemble of the upper stage, RV, and associated volumetric clutter such as debris is termed the threat complex. Ballistic missiles exhibit three kinematic phases corresponding to their flight regimes. The first is during boost when the missile is accelerating at least several g’s due to the axial thrust. This boost phase can employ several booster stages to accelerate the missile to a desired velocity and may include a low-acceleration stage for deployment of the RV. The second kinematic phase is when the RV or missile is flying a ballistic trajectory in the exoatmospheric. Only gravitational forces are significantly affecting the trajectory at this stage. The third phase initiates as the RV or missile reenters the atmosphere and drag forces contribute to shaping the trajectory. Both the boost and reentry phases contain acceleration and jerk terms that cannot be readily estimated a priori by the defense. The boost phase also imposes sudden changes in acceleration as burnout and staging events occur. Table 7-4 denotes representative details of the flight regimes for various ballistic range classes [4]. As indicated, the missile transitions from a relatively short boost period into ballistic motion through the midcourse regime. The altitude above about 100 km is commonly termed exoatmospheric to denote that atmospheric effects are

7.2 TABLE 7-4

¢

BMD Radar System Requirements

Representative Ballistic Missile Parameters

Ground Range (km)

Burnout Velocity (km/s)

Burn Time (s)

Burnout Altitude (km)

Apogee (km)

100 300 600 1,000 3,000 10,000

1.0 1.7 2.4 2.9 4.9 7.2

30–40 60–70 60–90 70–110 80–140 170–300

10–15 25–35 40–60 50–80 100–120 180–220

< 30 < 100 < 150 230 650 1,300

minimal while the region below that altitude is denoted endoatmospheric. Entry into the atmosphere introduces drag that diverts the missile from the nominal ballistic flight path as well as introducing significant aerodynamic and thermal stresses. Endoatmospheric entry also induces low-mass, high-drag volumetric clutter to ‘‘strip out’’ while potential threatening targets such as warheads continue on a ballistic trajectory. The accuracy of ballistic missiles is generally characterized by their circular error probability (CEP). The CEP is defined as the radius of the projected area within which the missile is expected to have a 0.50 probability of impacting. The CEP of modern ballistic missiles typically ranges from tens of meters to kilometers. The CEP decreases with decreasing guidance error and increasing mass–drag ratio. The improving accuracy of tactical ballistic missiles from advances in guidance and control technology significantly increases their potential lethality over the relatively inaccurate Soviet-era Scuds and third-party nation variants. While a number of currently deployed ballistic missiles utilize liquid fuel, there is an anticipated trend toward solid fuel systems as this latter approach decreases launch preparation requirements and can enable improved launcher mobility [5]. Ballistic missile payloads range from conventional high explosives to nuclear weapons. Tactical and theater ballistic missile may be designed to support multiple types of warheads, including chemical and biological weapons as well as submunitions. The intercept lethality of fragmentation warheads against this span of possible payloads is problematic. Achieving adequate lethality while avoiding the usage of interceptors armed with nuclear warheads has motivated the development and deployment of hit-tokill interceptors that depend on high closing velocities to effectively incinerate the ballistic missile payload upon collision. The kinetic energy released by high-speed collisions is significantly greater than could be achieved with an explosive warhead of equivalent mass to the interceptor. BMD radars identify and designate the warhead and RV to the interceptor to enable its endgame homing. There is typically a minimum acceptable intercept altitude specified for defense of ‘‘soft targets’’ such as population centers that drives the engagement timeline and battle space that must be supported by the BMD radar. Nuclear warheads must be intercepted sufficiently high to prevent their detonation from imposing damaging blast overpressure or thermal effects on the Earth’s surface. Chemical and biological payloads must be intercepted at sufficient altitude as to prevent their dispersion through the troposphere. In addition, interceptor design features may limit the minimum engagement altitude. BMD infrared seekers are largely confined to exoatmospheric operation due to sensitivity needs and atmospheric friction effects. Since the interceptor must possess sufficient

293

294

CHAPTER

7

Ballistic Missile Defense Radar

fly-out time to reach this minimum engagement altitude, modern BMD radars must acquire and track targets at long range in the exoatmospheric regime.

7.2.2 BMD Engagement Implementation The BMD battle space can be characterized as the spatial volume encompassing the earliest and latest intercept opportunity for a given threat scenario and operational context. In general, the engagement sensor should be sized to operate significantly beyond this battle space, which should be ideally determined by interceptor kinematic capabilities. The maximum intercept range increases with the commit range of the target at which track accuracy is sufficient to support launching the interceptor since the interceptor has more time for fly-out. Ballistic trajectories can be accurately predicted using high-fidelity representation of the equations of motion from extrapolating the track position and velocity. The extent of the defended footprint on Earth’s surface protected by the BMD system correspondingly increases with the fly-out time available to the interceptor. A co-located engagement sensor must operate over a significantly longer range than the interceptor to provide adequate fly-out time to support exoatmospheric engagements and defend large areas. As illustrated in Figure 7-2, the engagement radar will require sufficient time in track to achieve the prediction commit accuracy and discrimination confidence necessary for commit. Using the expressions given for this simple linearized example, assume self-defense against a ballistic missile attacking at 3,000 m/s with a 30 reentry angle targeted directly at the BMD site. The exoatmospheric interceptor is assumed to engage the warhead at an altitude of 100 km with an average fly-out velocity of 1,500 m/s. This scenario corresponds to an intercept range of 200 km with a time of flight of ~133 s. The corresponding commit range is 600 km due to the 2:1 velocity advantage of the threat missile over the interceptor. Allowing for 30 s time-in-track precommit, the engagement radar must initiate track at a range of 690 km against the threat missile. The required commit range will increase with increasing defended footprint extent to provide the interceptor additional fly-out time. BMD radars that provide early commit-quality tracks enable long fly-out times and large defended footprints. An early commit enables a large area, such as a nation, to be defended from long-range ballistic missiles using a single exoatmospheric interceptor

FIGURE 7-2 ¢ Simplified Engagement Geometry with Selfdefense Parametric Expressions.

Detection / Acquisition Timeline

Time in Track, Tt

Operational Ranges for Simplified Self-Defense (RFE = 0) Tfo = H/Visin(g) = Interceptor fly-out time Rc = (H/Visin(g)) (1+Vt /Vi) = Target Range at Interceptor Launch Rti = Rc+VtTt = Track Initiation Range

Commit Range, Rc Intercept Range, Ri Intercept Altitude, H Vi = Average Interceptor Velocity Vt = Average Warhead Velocity g = Reentry Angle of Ballistic Missile

g Forward Edge of Defended Footprint, RFE

7.2

BMD Radar System Requirements

site. Area defense motivates usage of an engagement sensor system that can achieve commit accuracy at the maximum target range congruent with the kinematic capability of the interceptor. The commit range can be extended by increasing the sensitivity and operational range of the radars, but this is ultimately constrained by radar line of sight (RLOS) and likely the incremental sensitivity cost. For example, the horizonconstrained range against an intercontinental ballistic missile (ICBM) with a ground range of 8,000 km is on the order of 4,000–5,000 km from the defended area. Forward basing radars nearer the ballistic missile launch area mitigates RLOS constraints and enables early acquisition and track. If there is an associated reduction in range requirements, forward basing enables usage of a smaller, less expensive radar. The AN/ TPY-2, which was developed originally to support theater missile defense, demonstrates the utility of a transportable radar system in this context. The engagement sensor system and resultant trend in forward-edge radius of the defended footprint, RFE, can be coarsely divided into two cases from system analysis considerations [6]. Both the time of flight and burnout velocity of a ballistic missile with flight path ground range GT are proportional to HGT. So, the potential fly-out time available to the interceptor increases with increasing ballistic missile ground range, but the corresponding threat velocity increases proportionately. It follows that we can parameterize BMD engagements into two classes such that: 1. effective sensor range significantly less than maximum threat ground range due to RLOS or sensitivity constraints – that is, RFE ~ 1/HGT corresponding to terminal defense with co-located radar and interceptors; and 2. effective sensor range enables early target detection and tracking corresponding to short-range theater ballistic missile defense (TBMD) scenarios or the use of networked forward-based radars for long-range fire control – that is, RFE ~ HGT. As indicated, this parameterization derives from considering the increase in burnout velocity and time-of-flight with increasing threat ground range as delineated in Table 7-4 in conjunction with a simple linear representation of the interceptor fly-out. The effective engagement sensor coverage and threat missile trajectory drive the achievable defended footprint. This case analysis motivates usage of forward-based engagement radars to counter long-range/high-speed threats that would otherwise contract the defended footprint. A BMD system track may require support from a network of sensors to cover midcourse as well as ascent phases of the threat missile flight to provide high-quality fire control. The canonical BMD engagement sequence and kill chain are depicted in Figure 7-3. The engagement sequence among BMD elements is depicted horizontally, while the functional sequence of the kill chain is depicted vertically in alignment with the corresponding BMD elements. The kill chain initiates with the autonomous detection of either launch events via launch-detection satellite (LDS) or ballistic missiles in flight via early warning and forward-based radars and subsequent reporting to the BMD command-and-control system. Metric and signature data from these sources are used to classify and characterize the threat. Ballistic missiles are tracked by multiple sensors with composite system track fidelity adequate to derive situational awareness, which includes launch-point estimation, impact-point prediction, and attack characterization. Using this tracking and classification data, the trajectories of threatening missiles are then predicted for handover to midcourse and terminal-engagement radars.

295

296

CHAPTER

7

Ballistic Missile Defense Radar

FIGURE 7-3 ¢ BMD Kill Chain and Engagement Sequence.

Surveillance Sensors

Engagement Sensors

Interceptors

Engagement Sequence Autonomous Detection Coarse Track for Situational Awareness // Cue Handovers Cued Acquisition // Seeker Designation Target Classification

Kill Chain

Precision Track for Fire Control Target Discrimination Kill Assessment

After achieving target acquisition, the engagement radars perform two critical functions – precision track and discrimination. Precision track consists essentially of tracking the desired target with sufficient accuracy to designate it to the kill vehicle (KV). This designation must be generated on the predicted trajectory and translated into Earth-centered inertial coordinate system accessible by the kill vehicle so that it is within the field of view of the seeker with high confidence. In parallel with the precision tracking, the engagement sensor identifies the primary target(s), discriminating the RV from other launch components, debris, or decoys using a combination of an extendedobservation metric and high-resolution signature data. Maintaining correct measurement-to-track association and well-resolved measurements is an implicit requirement for both the precision track and discrimination processes since these operations require an extended sequence of measurements for a given target. Modern BMD interceptors employ a separating KV that essentially consists of an infrared (IR) focal plane array sensor and some manner of lateral maneuvering capability. The postcommit midcourse phase of an exoatmospheric engagement consists largely of the interceptor flying to a point in space where the KV will be deployed from its single- or multiple-stage launcher. During the fly-out, the engagement radar may refine and uplink its track and discrimination estimates to the interceptor, taking advantage of both additional observation time and an increasing signal-to-noise ratio (SNR) due to decreasing range. A target object map (TOM) may be generated by the radar and uplinked to the interceptor to aid seeker acquisition of the designated target when multiple objects are assessed to be present in its field of view. The TOM is essentially the state vectors of the proximate target set mapped into seeker coordinates.

7.2

BMD Radar System Requirements

The KV autonomously conducts the endgame transitioning through target acquisition using the engagement sensor designation, target identification, homing track, and aimpoint selection [7]. The seeker acquires the target as the KV enters endgame and performs final target identification, terminal tracking, and aimpoint selection to maximize lethality against the target. Hit-to-kill KVs are designed to destroy the target through body-to-body impact as the high closing-velocity collision releases much more energy against the target than could be achieved with conventional fragmentation warhead for a given KV mass. The operation and survivability of the IR seeker mechanism as well as maneuverability constraints generally limit KVs to exoatmospheric intercepts. Interceptors have been developed to engage short-range ballistic missiles endoatmospherically but intercept of long-range ballistic missiles is generally conducted exoatmospherically. The determinism – corresponding to the absence of unknown accelerations – of an exoatmospheric trajectory enables ballistic target to be engaged by interceptors, which may only have a comparable or even lesser velocity than the target as long as they possess sufficient endgame maneuverability to correct for the radarprediction commit error and close out onboard guidance errors. The depicted overlap of the kill chain functions between engagement sequence elements in Figure 7-3 indicates that the function can be allocated to any element or shared among them for increased robustness. For example, if the engagement radar is challenged by sensitivity or resolution requirements, precision track and discrimination can potentially be off-loaded or shared with the KV. The national missile defense architecture that developed into the GMD element of BMDS initially allocated the kill chain functions solely between the surveillance sensor and interceptor elements. Under this concept, the existing early warning radars were modified to directly designate targets to the KV via the GMD fire-control and interceptor data-link elements mitigating the expense of developing and deploying new X-band engagement radars [8]. The reliable measurement performance, extended observation capabilities, and robust integration of acquisition and multiple-target track functions have ensured that radars remain as the principal BMDS engagement sensors. Enhancing resolution and accuracy performance as well as enabling transportability motivated development and deployment of X-band radars such as the SBX and THAAD radars for engaging ICBMs and theater ballistic missiles, respectively. Launch-detection satellites provide earlier detection but are incapable of precision tracking and attack assessment as they only observe the ballistic missile during boost. Future spaceborne sensors employing passive multiple-spectrum electro-optical sensors capable of tracking discrete objects will increasingly supplement this longrange surveillance mission. The generation of ‘‘stereoscopic tracks’’ can significantly improve accuracy by fusing angle-only measurements from multiple sensors with diverse viewing geometries to generate three-dimensional tracks. Even with this approach, electro-optical sensors will be limited in tracking and discrimination by their lack of range resolution and modest cross-range resolution. (Cross-range resolution can be approximated by the product of the sensor angular resolution in radians and the range to the target.) The complex of objects deployed from a single booster may contain intentional countermeasures designed to degrade BMD effectiveness as well as unintentional debris [9]. Countermeasures may include decoys of various degrees of sophistication ranging from simple balloons to replicas of the actual RV [10]. Other countermeasures

297

298

CHAPTER

Ballistic Missile Defense Radar

7

could be designed to mask the RV from BMD sensors by inducing volumetric clutter via chaff deployment or employing electronic attack. Finally, the RV design could be modified to distort or suppress key signature aspects, such as reducing RCS so as to degrade defensive radar effectiveness [11]. Discrimination of the RV from countermeasures and associated objects has long been viewed as problematic [12]. This challenge is particularly acute in the exoatmospheric midcourse regime where there are no atmospheric effects to separate heavy objects like RVs from light decoys. Entry into the atmosphere effectively serves as bulk filter against relatively lightweight countermeasures and debris. Some legacy BMD systems employed an overlay of long-range interceptors for high-altitude intercepts as well as an underlay of high-acceleration interceptors for endoatmospheric engagements of RVs that had leaked past the initial battle space or had been protected by countermeasures until reentry. However, defense of large areas and soft targets, such as population centers, mandates interceptor launch well before reentry in order to enforce highaltitude intercepts and defend large areas on Earth’s surface. Unless the BMD radar can reliably discriminate the RV from other objects, the BMD system must incur the wastage of committing an excessive number of interceptors against a given ballistic missile complex. Conversely, if the BMD system has a limited interceptor inventory, leakage will occur when the RV is not engaged due to incorrect discrimination or inaccurate designation to the KV.

7.3 RADAR DEVELOPMENT FOR BALLISTIC MISSILE DEFENSE The history of BMD with focus on radar aspects is summarized in Figure 7-4. While much attention has focused on successive generations of interceptor technology progressing from exoatmospheric and endoatmospheric missiles armed with nuclear FIGURE 7-4 ¢ BMD Development Timeline.

US Programs

Nike-Zeus

SU/FSU Programs

Sentinel

Safeguard

LOADS

ABM Treaty Initial Moscow BMD

Research Population Defense Site Defense TMD Deployment Population Defense Site Defense TMD Interceptor Technology UHF Early Warning Radar Clear, AK Thule, Greenland Fylingdale, UK

TIR

GBR

BMDS Blocks

SDI Speech Moscow Upgrade

Command-Guide Nuclear

ABMT W/D

IR-Seeker Hit-to-Kill SS PAR SS PAR SS PAR

S-band X-band Phased Array

ABM Radar Technology

Special SP & DP

1960

1970

TWT COTS DP & Special SP

1980

1990

Solid State COTS SP/DP

2000

7.3

Radar Development for Ballistic Missile Defense

warheads to the present emphasis on kinetic kill vehicles, the progressive development and deployment of unique radar systems has been critical to advancing capabilities. Early research supporting BMD efforts was critical to developing the first generation of phased array radars for air defense. Later BMD research was critical to the emergence of AESA and high-throughput processing technologies. The large sunk cost required by this class of radars strongly motivates spiral development where new capabilities are inserted largely via new software builds as well as processing and communications hardware upgrades. The high cost of upgrading the radio-frequency (RF) apertures of large phased array radars has historically precluded that option. The primary source of situational awareness data for BMDS is the large phased array radars that were initially designed and deployed to support Cold War MW but later upgraded to support BMD. The COBRA DANE radar initiated operations in 1977, and the UHF early warning radars were developed in the 1980s. In addition to utilization of advanced technologies such as solid-state phased array radars, their operational sizing and attendant physical extent is also a distinguishing characteristic of BMD radar systems. As defined in Chapter 6, the noise-limited search capability of a radar system is ultimately determined by the product of its average transmit power and receive antenna aperture dived by the product of its system noise temperature and system loss factor. BMD search requirements lead to the necessity for large antenna aperture fed by high average power. U.S. and Russian early warning radars are integrated into the sides of multiple-story buildings that also house the operations center, transmitter, receiver, and processing subsystems. Even TBMD radars designed to counter short-range theater missiles are large in comparison to their air-defense counterparts. For example, the X-band THAAD radar system antenna was designed to be transportable but is more than an order of magnitude larger and several orders of magnitude more capable than X-band AESAs used in airborne-fighter radars [13]. The modern generation of BMD radars addressed in this chapter is somewhat arbitrarily defined as those employed since the cessation of the Cold War in the 1990s, but all of the systems in use now were designed or initially deployed during that period under the constraints of the Anti-Ballistic Missile (ABM) Treaty as addressed shortly. Current BMD radars consist primarily of multiple-function phased array radar systems that were originally built to support early warning and attack assessment, TBMD, and area air defense. MDA is developing the BMDS composed of sensors, weapons, and command-and-control networks intended to provide increasing defense capabilities over a series of block upgrades and phased deployments as adversarial offensive capabilities grow and defensive technologies mature. The objective goal of the BMDS is provide the global capability to engage ballistic missiles of all ground-range classes throughout their entire flight regimes. The ABM Treaty of May 1972 sought to ensure the stability of strategic competition based on mutually assured destruction such that neither the United States nor the Soviet Union (USSR) could instigate a first strike without fear of suffering a devastating retaliatory strike from the adversary’s surviving strategic forces. Toward this end, the ABM Treaty limited deployment of BMD systems to the defense of two areas for each country; this was later amended to a single protected area. The USSR chose to develop a BMD system to defend Moscow, while the United States developed and briefly deployed a system to defend ICBM launch complexes. The United States activated the Safeguard BMDS system in North Dakota in 1975, but it was shut down within months.

299

300

CHAPTER

7

Ballistic Missile Defense Radar

This system employed the UHF Perimeter Acquisition Radar (PAR) and S-band Missile Site Radar (MSR) to support an overlay defense of exoatmospheric Spartan interceptors and an underlay defense of endoatmospheric Sprint interceptors. The PAR and MSR were both multiple-function phased array radars employing tube-based transmitters driving passive phased array antennas integrated into hardened building structures. Both classes of interceptors employed tailored nuclear warheads so that radar-uplinked command guidance provided adequate engagement accuracy. Perceiving value in a limited defensive capability for their national capital regions, the Soviet Union deployed and later upgraded a Moscow BMD system that remains operational in current-day Russia. The principal constraints imposed by the ABM Treaty and subsequent interpretations of it on radar system and technology development are: ●

bans on developing, testing, or deploying ABM systems or components that are ‘‘seabased, air-based, space-based, or mobile land-based’’; constraints on deployment of future early warning radars to the periphery of national territory with outward orientation; implicit definitions of ABM radar as phased array having average transmitted power – aperture area product in excess of 3 million; and interpretations that ban testing of air-defense components in ‘‘ABM mode,’’ including radar tracking of ‘‘strategic ballistic missiles’’ or operating ‘‘in conjunction with the test of . . . an ABM radar at the same test range.’’

As delineated in Chapter 6, the search form of the radar-range equation indicates that noise-limited performance is proportional to the product of average transmit power and receive-antenna aperture area. The implicit definition of an ABM/BMD radar in terms of average power – aperture area product in excess of 3 million w-m2 – appears motivated by estimation of the search capability required to perform the strategic BMD surveillance mission along with assumed system noise temperature and loss characteristics likely based on the technologies of the time. The ABM Treaty is the only international agreement known to the author that makes implicit reference to the radar-range equation. Under the ABM Treaty, BMD system development efforts in the United States subsequent to Safeguard were largely directed toward defense of strategic force assets in recognition of the technical challenges of defending dispersed and highly vulnerable population centers in contrast to hardened ICBM launching facilities and associated command-and-control installations. Research efforts produced the fixed S-band Site Defense Radar followed by the transportable X-band Sentry radar system concepts. Both radar systems were intended to support low-altitude engagements against counterforce attacks on hardened ICBM sites using high-acceleration command-guided interceptors intended to conduct endoatmospheric engagements. These radars intended to take advantage of the endoatmospheric battle space to ease the discrimination requirements and aggressively utilized nuclear hardening technology to enable survivability in the event of conflict. The associated BMD interceptors all employed command guidance with nuclear warheads, which imposed a significant lethality radius but presented the operational and technical challenges of operation in a nuclear environment. Substantial effort was expended on research and development, but the associated BMD systems

7.3

Radar Development for Ballistic Missile Defense

were never fabricated and deployed due to concerns over cost, effectiveness, and impacts on strategic stability. President Ronald Reagan announced in March 1983 that he was initiating development of a BMD system that would provide high-confidence protection of the entire United States population. This system was anticipated to make extensive use of spacebased sensors and weapons to provide global protection against ballistic missiles, leading to the appellation of ‘‘Star Wars’’ for this conceptual system. The anticipated transition from ground-based radar systems to spaceborne electro-optical sensors encountered a number of technical and affordability challenges leading to the present emphasis on spiral development of legacy radar systems to support BMDS and continued research efforts to develop spaceborne electro-optical sensors. In parallel with the increased interest in radar for strategic BMD, Iraqi military operations using modified Scud theater ballistic missiles (TBMs) motivated new interest in TBMD. Research and demonstration efforts led to ongoing modifications in the U.S. Patriot and Aegis air-defense systems to detect, track, classify, and engage TBMs. The Russians also developed extended air-defense systems such as the S-300(V), which incorporated upgraded radars and interceptors to destroy TBMs. While regarded by some as the cornerstone of strategic arms agreements, the ABM Treaty significantly constrained the development and deployment of radar systems with BMD-like characteristics. Upgrades of early warning systems and the development of mobile TBMD systems were, in principle, subject to bilateral adjudication to ensure compliance with the treaty. U.S. efforts to deploy a national missile defense appeared significantly constrained by the treaty. The constraints on U.S. BMD development led the United States to abrogate the ABM Treaty in June 2002. However, all the American and Russian BMD radar systems currently deployed were developed under its constraints. The great achievement of the BMDS effort is arguably accomplishing strategic missile defense with sensors that were designed expressly under treaty obligations to not possess this capability. Through innovative BMD system architectures, modern networking constructs, and advanced interceptor development, these radars support military utilities far beyond their original design intent. Advancing the timeline of launch detection is precious to tactical attack warning as well as to BMD. The United States developed the Defense Support Program series of launch-detection satellites, which used infrared sensors to warn of ICBM and SLBM launches. Recent Russian writings indicate that the Russian LDS program was far less successful than that of the United States, leading to the deployment of over-the-horizon radar (OTHR) systems for early warning as well as construction of a network of large phased array radar systems around the periphery of the USSR to provide attack assessment and support their BMD system. The United States did not deploy OTHR systems for early cueing likely because of concerns over data reliability and availability constraints imposed by solar events and diurnal cycles on ionospheric propagation performance.

7.3.1 U.S. BMD Radar Deployment For decades, the United States has maintained a network of large UHF radar systems sited and oriented to provide tactical warning and attack assessment of ICBM attacks on the United States. This mission imposes search and track sizing requirements that also

301

302

CHAPTER

7

Ballistic Missile Defense Radar

make these radar systems well suited to support BMD operations. Upon deactivation of the Safeguard BMD system, the PAR surveillance sensor near Cavalier, North Dakota, was converted into an attack assessment MW role as the Perimeter Acquisition Radar Characterization System (PARCS). However, since PARCS is located in the northern United State, it cannot provide early warning due to its horizon-limited range constraints. The United States deployed the Ballistic Missile Earn Warning System (BMEWS) radars at locations near Clear, Alaska; Thule, Greenland; and Fylingdales Moor in the United Kingdom. These radar systems were sited to provide for the detection and assessment of Soviet ICBM attack on the United States. The initial deployments consisted of an ensemble of UHF switched-fan-beam radar systems and mechanically scanned single-target tracking radars. The BMEWS deployment was supplemented by a series of mechanically scanned radars sited to detect SLBM launches against the continental United States. This original BMEWS deployment circa 1960 was deemed inadequate due to incomplete coverage of possible Soviet attack trajectories, coarse resolution, inadequate accuracy, and limited target-handling capacity. The BMEWS and primary SLBM detection systems were all eventually upgraded to multiple-faced UHF multiple-function AESA radars integrated into the building structure that houses operations and supporting systems. There are currently two active PAVE PAWS radar systems sited in Massachusetts (Cape Cod Air Force Station) and north California (Beale Air Force Base). While these radars were developed and sited to address the Soviet threat to the United States, they have been adapted to provide coverage against emerging threat states as depicted in Figure 7-5. The BMEWS and PAVE PAWS operate in the 420- to 450-MHz UHF band and employ circular polarization according to spectrum allocation data. The PAVE PAWS each use two transmit-and-receive apertures covering a composite azimuth extent of 240 . Each 22-m aperture possesses about 1,800 active elements reportedly generating about 340 watts each of peak RF power [14]. The BMEWS radar apertures possess about 2,560 elements each and are some 26 m across [15]. While the BMEWS radars at Thule, FIGURE 7-5 ¢ U.S. BMD and MW Surveillance Radar Coverage. Fylingdales Thule Clear

Bealee

Cape Cod

7.3

Radar Development for Ballistic Missile Defense

303

Greenland, and Clear, Alaska, are two-sided, the one near Fylingdales Moor in the United Kingdom employs three RF apertures for 360 coverage. These arrays are thinned designs that sacrifice aperture efficiency and sidelobe control in exchange for decreased beamwidth. This design also enables potential sensitivity growth by fully populating the RF aperture with active elements. These radars are sized to detect and track threat complexes at the maximum RLOS range of beyond 4,000 km against longrange ballistic missiles. COBRA DANE was activated in 1977 on Shemya Island, Alaska, and has performed key roles in missile warning, BMD, and SSA as well as enforcing strategic arms control treaties by monitoring Russian missile testing. This multiple-function phased array radar employs 96 traveling wave tube transmitters distributed across the 30-m diameter corporate-feed antenna. The resulting spatial power combining enables very high peak power operation on the order of 15 MW and an average transmit power of 0.9 MW [16]. This L-band system operates in the 1,215- to 1,400-MHz band. COBRA DANE supports a wideband linear frequency modulation (LFM) waveform of some 200-MHz extent, which should provide a range resolution of about 1 m, taking into account broadening from weighting to suppress range sidelobes. The Sea-Based X-Band (SBX) radar system utilizes a limited field-of-view (LFOV) AESA on a mechanical mount to provide adequate field of regard as illustrated in Figure 7-6. The ensemble is protected by a radome. About 65 percent of the 384-m2 aperture is populated with active transmit-and-receive modules driving radiating horn elements. The X-band radar system is mounted on a self-propelled semisubmersible platform that was originally designed to support oil drilling. The platform measures 73 m wide by 118 m long and contains the power plant, control room, living quarters, and other infrastructure to support the SBX mission. The similar but smaller GroundBased Radar–Prototype (GBR-P) at the Ronald Reagan Ballistic Missile Defense Test Site on Kwajalein Atoll in the Marshall Islands was used to retire the associated risk FIGURE 7-6 ¢ SBX Radar and Platform.

304

CHAPTER

7

Ballistic Missile Defense Radar

with discrimination, electromechanical scanning, target object map generation, and hit assessment as well as support of GMD flight testing. The AN/TPY-2/THAAD radar shares a significant degree of common technology with the SBX radar but is a full field-of-view design. Both radar systems were developed by Raytheon; this firm also constructed the BMEWS, PAVE PAWS, and COBRA DANE radar systems. The THAAD radar is an X-band AESA with a rectangular 9.2-m2 antenna aperture supporting 25,344 transmit–receive modules [17]. This radar performs surveillance, tracking, and discrimination against tactical and theater ballistic missiles in support of BMDS as well as the THAAD TBMD system. The radar system is transportable via either C-5 or C-17 and consists of the antenna unit, electronics unit housing the receiver and digital-processing subsystems, cooling and power distribution unit, and a prime power diesel-generator unit.

7.3.2 Russian BMD Radar Deployment The Russians began development of their BMD systems with supporting large groundbased radar systems during the 1960s. Their first-generation BMD system defending the Moscow region included mechanically scanned fire-control radars as well as large UHF battle-management and early warning radar systems [18]. These two Dunay-class battlemanagement radars were located in Moscow and known to Western analysts as Dog House and Cat House. They reportedly provided a 3,000-km detection and tracking range against RVs using frequency-scanned continuous wave techniques [19]. The Russians developed several variants of VHF azimuthally frequency-scanned early warning radars, including the Denestr, Dnestr-M, and Dnepr systems featuring antenna apertures approximately 200 m wide by 20 m tall. These radar systems were known collectively in the West as the Hen House series. These were reportedly multiple-face sites to achieve the requisite azimuth coverage. The UHF Volga variant of this architecture was intended to provide surveillance against intermediate range ballistic missiles and SLBMs [20]. Beginning in the 1970s, the USSR began deployment of the large Daryal-series VHF phased array systems featuring separate transmit and receive arrays in independent structures separated by 2–3 km. As depicted in Figure 6, these systems were known in the West as the Pechora-class Large Phased Array Radars, named for the location of an early installation. These systems reportedly employ phase-coded extended-duration waveforms and advanced signal-processing and electronic-protection features. Their operational flexibility is sufficient to support a mixture of early warning, BMD system support, and space surveillance. A significant innovation of these radars is that either the transmit or receive array could be reportedly paired with an existing Hen House frequency-scanned radar system, though obviously with degraded performance from the full dual-aperture configuration. These systems were physically massive, with the Krasnoyarsk installation being described as possessing a 30-story radar receiver and an 18-story transmitter [21]. These systems likely provided superior target handling and tracking accuracy to the previous generations of frequency-scanned arrays, but their relatively low frequency of operation likely imposed operational limits due to ionospheric propagation effects and electromagnetic interference. Several of the key Soviet-era early warning radars were emplaced in territory that became independent nations with the end of the Cold War, resulting in their subsequent

7.3

Radar Development for Ballistic Missile Defense

destruction or operational compromise. As a result, the Hen House network continued to shoulder onward decades past its initial deployment. Podvig lists ten surviving early warning radar sites, with some supporting several radar systems of different generations or azimuthal coverage [22]. At the peak of the Soviet early warning radar, these systems collectively provided almost total coverage of potential ballistic missile attack trajectories. Recent Russian press accounts have announced deployment of the new class of Voronezh early warning radars. The Voronezh radars can reportedly be rapidly constructed from factory-manufactured modular components. These radars are also reportedly far less expensive to construct and operate than predecessor systems. Efforts are currently underway to develop and deploy a new generation of surveillance radar systems around the periphery of Russia to support missile warning, missile defense, and SSA. Russian sources have claimed maximum tracking range against ballistic targets and satellites of some 6,000 km from their early warning network radars. In parallel with the VHF line-of-sight early warning radars, the USSR also developed and deployed sky wave over-the-horizon radar capability that reportedly would provide launch-detection capability against U.S. ICBMs [23]. This work may have been motivated by the Soviets’ reported difficulties in developing and operating launchdetection satellites. The OTHR effort was not without setbacks either: One site was reportedly located next to the Chernobyl nuclear plant. It is likely that the sky-wave propagation channel necessary to detect a U.S. ICBM launch suffers frequent outages as the radar must operate though the polar ionospheric region. The Daryal-class LOS early warning radars would have been crucial in detecting and assessing a strategically threatening ballistic missile attack that might have been launched against the USSR and likely did prevent military escalation from false attack indications that arose from less reliable sensors. The early warning radar network and the Dunay battle-management radars supported the mechanically scanned fire-control radars of the initial Moscow BMD system termed the ‘‘A-35’’ [24]. The system reportedly employed command-guidance interceptors armed with high-yield nuclear warheads. The A-35 reportedly achieved some measure of operational capability in 1972 but with limitations well recognized by the Soviet leadership. It appears to have been extremely limited in its ability to accommodate multiple engagements and to mitigate countermeasures. The USSR in 1989 reportedly developed and deployed the successor A-135 BMD system to defend Moscow with an initial operational capability as illustrated in Figure 7-7 [25]. This system employed both exoatmospheric overlay and endoatmospheric underlay interceptors. Battle management and fire control are integrated into the multiple-function Don-2N radar system deployed north of Moscow near Pushkino. This radar system is built into a four-sided truncated pyramid some 40 m tall with equal length and width sides of some 100 m extent [26]. Each face reportedly has a circular phased array aperture about 16 m in diameter and a square phased array aperture of about 10 m extent. The radar reportedly operates in the mid-microwave band and possesses significantly improved resolution capabilities over the predecessor system in addition to multiple target-handing and electronic-protection features. This radar has reportedly tracked 5-cm-diameter spherical satellites at some 2,000-km range [27].

305

306

CHAPTER

7

Ballistic Missile Defense Radar

FIGURE 7-7 ¢ Russian Don-2N (‘‘pillbox’’) Radar.

Like their counterparts in the United States, Russian BMD developers probably were challenged by discrimination of targets from debris and countermeasures. Some recent Russian writings suggest that the Soviets may have attempted to address this challenge by utilizing a precursor high-yield nuclear burst to destroy or mitigate low-mass countermeasures [28]. The resultant deviation in target-trajectory and target-polarization characteristics would reportedly have served as a basis of discrimination for the fire-control radar. However, this approach would have mandated that the radar operate in a nuclear environment, thus imposing severe challenges.

7.3.3 International BMD Radar Deployment The enormous investment of resources required for BMD against ICBM attack has limited research, development, and deployment to the United States and Russia to date. However, the diffusion of ballistic missile and weapons of mass destruction technology has given regional peer competitors the capability to strategically threaten each other with theater ballistic missile weapons. In addition, the increasing accuracy of TBMs and the availability of more lethal conventional warheads have markedly increased their significance in tactical warfare. There will likely be increasing interest in warning sensors and BMD radar systems capable of supporting defense against TBM attack but within the budgets and operational capabilities of emerging regional powers. A number of radar systems offering some degree of TBMD capability have been developed, deployed, or purchased in the international arena. There appears to be significant interest in upgrading exported Aegis and Patriot systems to support TBMD. Citing legacy development efforts to counter TBMs, Russia is offering updated area airdefense system variants with TBMD capability such as the Antei-2500. Israel developed the Arrow TBMD system, which includes the L-band Green Pine multiple-function AESA [29]. The new generation of European multiple-function radar systems developed to support area air defense may be upgraded to support TBMD operation in conjunction with weapon system enhancement. India and Japan are integrating internationally procured and domestically developed elements to develop BMD capabilities.

7.4

7.4

BMD Radar Design

307

BMD RADAR DESIGN

7.4.1 Frequency Considerations As a general rule, the United States has historically tended to choose higher bands for BMD radars than optimal from consideration of target RCS and cost-constrained noiselimited sensitivity. This trend has been motivated by the combination of mitigating propagation effects, the desire for enhanced resolution to mitigate the effects of dense object environments and countermeasures on the engagement process, and enhanced track accuracy for reliable interceptor guidance. The USSR chose VHF for its early warning and battle-management radar systems, which is nominally optimal for detection as RV-sized targets are near their resonant peak RCS value. Small debris are in the Rayleigh region, resulting in decreased RCS so as to suppress volumetric clutter returns. The resonance enhancement and Rayleigh suppression RCS regions are both easily evident in Figure 7-8, which depicts the normalized RCS of a conducting sphere. (In comparison, microwave radars such as those used as engagement sensors and in forward-based surveillance operate in the optical RCS region against typical targets.) U.S. designers chose UHF for surveillance radars to mitigate ionospheric propagation degradation in the natural environment and severe blackout effects anticipated in nuclear environments while retaining some low-band advantage in target RCS enhancement and debris suppression. In addition, the UHF radar spectrum generally suffers less severe electromagnetic interference than often encountered at VHF. Similarly, the United States migrated from S-band to X-band for BMD engagement radars to improve operation in a nuclear environment and to achieve finer range and range–rate resolution for enhanced discrimination. Moreover, interest in developing transportable BMD radars motivated interest in X-band as higher frequency enables a Radar Cross Section of Metal Sphere with Asymptotes

Monostatic RCS / Projected Area · Asymptotes

5

FIGURE 7-8 ¢ Normalized RCS of Conducting Sphere.

1 5

–1

10

5

10–2 5

10–3 5 10–1

2

4

1 6 8 1 2 4 6 8 10 2 Relative Frequency = Circumference / Wavelength

4

2

6 8 10

308

CHAPTER

7

Ballistic Missile Defense Radar

smaller antenna aperture for a given level of sensitivity and tracking accuracy. While AESA technologies promise the prospect of increasing the technically achievable bandwidth to at least 20 percent of the carrier frequency, the usable bandwidth below X-band is tightly constrained by spectrum allocation regulations. Above X-band, the cost of high-power RF generation increases markedly as does atmospheric attenuation. Hence, X-band has come to be viewed as the most desirable frequency band of operation by the U.S. BMD community. Affordability of large X-band BMD-class radar systems is a continuing challenge. Radar operation in a nuclear environment is required if nuclear busts occur near surveillance zones or battle space. The ionized region of the fireball and immediately surrounding atmosphere tend to be relatively small. However, beta radiation emitted by the radioactive debris of high-altitude nuclear bursts can raise ionospheric electron densities by orders of magnitude over an extended region. The D-region of the ionosphere centered at an altitude of about 65 km has been assessed as exhibiting the most significant prolonged effects [30]. Ionization effects can be parameterized on the basis of the corresponding freeelectron density, which determines the plasma frequency. RF waves at frequencies below this value are reflected from regions of elevated ionization. RF waves above this frequency penetrate but suffer attenuation that decreases with the increasing square of the frequency. The effects of this total reflectance or high degrees of attenuation on radar operation is termed nuclear blackout as it masks the surveillance volume and battle space beyond the ionized region [31]. BMD radar system designers may mitigate nuclear environmental effects by designing their systems at the highest practical radar frequency for the specified functionality. However, even X-band systems will be subject to attenuation, refraction, and phase dispersion in regions of high ionization.

7.4.2 Implementation 7.4.2.1 Search and Acquisition Surveillance to achieve autonomous detection is typically performed by long-range VHF or UHF radar systems such as previously noted. Depending on capability and siting, these long-range radars may also perform battle-management functions generating handoffs for engagement radars and providing situational awareness tracks. Early warning against ICBMs mandates some manner of forward basing to mitigate RLOS limitations on detection range. Radars located within national boundaries can offer surveillance against SLBM or TBM attacks as well as support attack characterization, long-range interceptor engagement support, and target designation to engagement radars. However, the range resolution at these bands is inadequate to reliably resolve CSOs or discriminate individual targets. Autonomous search for exoatmospheric ballistic missiles mandates large radar systems with capabilities far exceeding that encountered in air-defense applications. For example, published technical data indicate that the PAVE PAWS possesses an average power – aperture product of about 6 107 w-m2, neglecting aperture efficiency and other losses [32]. While meeting the ABM Treaty threshold of BMD capability, this system is sized for SLBM detection, which demands less sensitivity than required by BMEWS to detect and characterize ICBM attacks due to the shorter range of operation. In addition to these noise-limited sensitivity demands, modern BMD AESA systems must jointly support high duty cycle and large range windows, which challenges

7.4

BMD Radar Design

timeline occupancy. The radar planning and scheduling process must interleave search and track processes within the timeline required by mission requirements. Based on a priori knowledge of threat launcher location and ballistic missile flight characteristics, a series of search fences may be devised that enable the radar system to focus its surveillance resources along anticipated threat trajectories. Targets detected in a search fence can then be tracked throughout the radar’s field of view. In addition to supporting early target detection, this technique also enables efficient timeline usage. A surveillance radar system can cue an engagement radar system, both extending its track-initiation range and increasing its target-handling capability by offloading autonomous search operations. Cueing from broad area sensors such as space-based infrared sensors can potentially extend radar acquisition and track-initiation range [33]. As noted previously, increasing track-initiation range can extend the intercept commit range and hence the extent of the defended footprint. In practice, the propagated error from a coarse cue may demand substantial search capability on the part of the engagement radar. The military utility of tailoring the placement, extent, dwell sensitivity, and revisit rates search fences for specific autonomous operation and cued acquisition applications motivate the usage of multiple-function phased array radar systems for surveillance. In addition to beam agility, these systems typically employ energy-management techniques in their control process so that individual beam-position transmissions are of durations appropriate for a target class of specified RCSs at a corresponding maximum range.

7.4.2.2 Tracking and Discrimination BMD imposes unique tracking challenges, including high degrees of measurement-error smoothing, multiple target tracking in the presence of closely spaced objects, and registration in an absolute coordinate system. Successfully meeting these challenges requires specialized radar design features such as high-resolution tracking and enhanced calibration and alignment techniques as well as sophisticated algorithms for measurement-to-track data association and track filtering. The tracking and discrimination processes must be closely integrated. Any robust discrimination process requires multiple measurements over an extended interval on a given target, mandating firm track as a necessary condition. Having expended radar timeline and energy on target to identify the target, it must be kept in track to support subsequent designation to the interceptor. To attain the accuracy required for midcourse intercept support, BMD engagement radars must and do attain significantly superior track-filtering performance than that of sensors tracking maneuvering air-breathing targets. Given that gravitational forces are the only significant effectors on target trajectory in the exoatmospheric regime, the track filter can be designed to attain a very high degree of smoothing. A six-state Kalman filter or batch-processing technique optimized to exploit the low process noise and replicate equations of motion is appropriate for this purpose. The states under estimation are simply the position and velocity vectors. During reentry, a Kalman filter with an additional state to account for drag is generally utilized. Nine-state filters with terms to account for more complex aerodynamic forces may be used to optimize tracking performance for a maneuvering reentry body or other target anticipated to exhibit significant lift or aerodynamic control capability. A nine-state filter with an acceleration vector in addition to position and velocity for state estimation may be employed against boosting targets. However, tuning the filter

309

310

CHAPTER

7

Ballistic Missile Defense Radar

plant noise (error model) terms for targets that transition almost instantaneously between boost and coast states is problematic and imposes significant performance compromises. The interacting multiple model (IMM) track filters performs well in this application [34]. This technique incorporates parallel state models, updates them in parallel with new measurements, and then blends the composite output and update variables. The IMM filter can accommodate boost and coast state model representations that accommodate booster burnout [35]. Successful track filtering presupposes that measurements of sufficient quality have been successfully extracted from the target under track. Data-association performance decreases with increasing update intervals, target maneuvering, and measurement error. The challenge of correctly associating measurements with the proper target in a CSO environment has been well documented in the academic literature. This problem can be particularly severe in an exoatmospheric environment since there is no drag to dissipate concentrations of CSOs. While exoatmospheric track filtering can achieve excellent smoothing performance the lack of dissipative forces enables the persistence proximity of CSOs challenging the capability of observing sensors to produce individually resolved measurements. BMD radars must resolve CSOs in range or Doppler to support the tracking and discrimination processes.

7.4.3 BMD Radar Technologies 7.4.3.1 RF Aperture The RF aperture can be defined to include the antenna, integrated RF components, beam-forming network, and structurally integrated mechanical and electrical components. Beam agility is required to support integrated search and multiple-target track. The primary performance challenges imposed on RF aperture technology by BMD applications are sensitivity, bandwidth, and metric accuracy. Sensitivity denotes the need to achieve an adequate SNR and measurement rate on small targets at long ranges. The need for fine range resolution drives the RF aperture to be designed for wideband operation since the inverse of the waveform bandwidth bounds resolution in the range dimension. Designation of discriminated targets to the BMD battle manager or weapons system requires precise tracking. The requisite accuracy must often be attained in an absolute coordinate system so that coordinate misregistration can be a dominant BMD system track error source. From the perspective of the acquisition community, cost is the driving concern typically associated with BMD radar systems. Modern BMD radar systems employ RF apertures constructed using AESA technologies as opposed to centralized or tube-based transmitters driving passive phased arrays or mechanically scanned antennas. The AESA RF aperture includes supporting power conversion and conditioning subsystems, beam-steering processor, and cooling and mechanical support structures in addition to the AESA transmit–receive module (TRM) assemblies and radiating elements. The RF aperture can impose in excess of 70 percent of the recurring cost, which is typically several hundred million dollars for a modern BMD radar system. The inherent redundancy and anticipated low component-failure rates of AESAs in surveillance and engagement radars are critical to overall BMD system reliability. The noise-limited sensitivity that determines a radar’s capability to collect measurements in support of tracking and discrimination is determined by the product of

7.4

BMD Radar Design

average transmit power and the square of the antenna aperture area for a given frequency, system loss, and system noise temperature as delineated in the track version of the radar-range equation. Through a simple application of optimization theory, it can be shown that this relationship dictates that the aperture should cost twice as much as the transmitter to maximize performance for a given total cost. Conventional AESA architectures are generally biased toward power relative to the optimal design point. AESA architectures that implement lower power density than currently practiced promise reduced acquisition costs but impose larger antenna apertures. While a single TRM could be employed to drive multiple elements to achieve low power density, this would degrade the low-loss transmission and reception path afforded by AESA technology. Both American and Russian developers have investigated usage of LFOV electronically scanned arrays to reduce RF aperture cost. Hybrid systems that incorporate both electronic scanning and mechanical scanning provide expanded field of regard at the cost of increased mechanical complexity and restricted transportability associated with the requisite antenna mounts required to accommodate the mass and size of BMDclass RF apertures. The American SBX/XBR design utilizes high-gain horns as radiating elements so that the number of modules is reduced by more than an order of magnitude for a given aperture area. As addressed in Chapter 6, an ESA is restricted to scanning its beam within its element pattern, which decreases with increasing element gain and thus limits the field of view. BMD radars may have to survive electromagnetic pulse (EMP) effects originating from high-altitude nuclear bursts and operate in an environment when such a burst has occurred. A high-altitude electromagnetic pulse (HEMP) can induce a sufficient level of EM energy to disrupt or destroy electrical systems and electronic components over a large surface of potentially thousands of miles extent [36]. HEMP bursts would occur above 30 km so that there is no appreciable blast or thermal effects at Earth’s surface, eliminating the need for hardening modern BMD radars to these effects. The gamma radiation from such a burst induces a large-scale disposition of electrons across a broad extent of hundreds to thousands of km of the upper atmosphere, inducing EMP generation. While only a small portion of a nuclear burst is converted into EMP, this can result in very-high-peak electric field intensity on Earth’s surface. A high-altitude burst of high yield could impose peak intensity of 100,000 v/m [37]. The EMP possesses multiple phases occurring over a time scale of microseconds to a second. Most of the energy would be concentrated at low frequencies relative to those used for LOS radar operation. While HEMP is a national concern due to the potential vulnerability of unprotected communication and power grids as well as specific devices, hardening and mitigation techniques applicable to radar systems and supporting subsystems are well established though potentially expensive.

7.4.3.2 Signal and Data Processing As described previously, the exoatmospheric BMD environment does not impose the drag effects that eliminate volumetric clutter upon reentry. Identification of threatening objects from decoys and debris is problematic under these conditions motivating interest in high-resolution measurement techniques. Wideband linear frequency modulation on pulse is an attractive waveform for high-resolution BMD applications. The Doppler tolerance of LFM is attractive given the long pulses that may be required for adequate sensitivity and the large mean and spread of target range-rates. LFM supports active correlation or stretch processing whereby the pulse compression is effectively divided

311

312

CHAPTER

7

Ballistic Missile Defense Radar

between an initial down-conversion with an LFM reference signal matched to the transmitted waveform within the receiver followed by a Fourier transform in the signal processor to provide range-gated outputs across a broad range swath. In addition to decreasing analog-to-digital conversion bandwidth demands by several orders of magnitude, this technique also significantly decreases signal-processing throughput and memory requirements. BMD radars employ an ensemble of waveforms of different bandwidth and durations that are scheduled by a real-time resource management process according to operational needs and radar capabilities. Like the radar community at large, BMD radar developers typically rely on commercial-off-the-shelf signal and data processing subsystems. BMD tends to be a high-throughput application space compared to the larger radar community. The usage of high-resolution waveforms over large range windows motivates significant signal-processing throughput and memory sizing. The potential presence of volumetric clutter also motivates sizing the computational capacity of the radar data processor to accommodate large numbers of simultaneous tracks. The long operational lives and evolving missions of BMD radars should strongly motivate investment in modular open system architecture practices for software to enhance software maintenance, mission growth, and software/hardware platform rehosting.

7.5

BMD RADAR PERFORMANCE ESTIMATION

BMD radars are often designed and evaluated in terms of their noise-limited performance. This practice is in contrast to air-defense applications where performance is often determined by the radar capability to detect small targets in the presence of strong ground clutter. The emphasis on optimizing the radar design for noise-limited performance is a consequence of the BMD battle space typically being well beyond the maximum clutter range due to surface or atmospheric volumetric returns. This condition may not be met for short-range BMD applications in geographical areas where highrelief surface features or anomalous propagations are present. Electronic protection techniques such as sidelobe cancellation or sidelobe blanking may be required to mitigate intentional and unintentional electromagnetic interference. The noise-limited search and track performance of BMD radars can be roughly estimated using the radar-range equation metrics presented in Chapter 6 but high-fidelity dwell-by-dwell simulation is required to fully assess performance. It is imperative that simulation tools adequately represent operational and environmental loss sources that impact radar performance. The extended range windows required for BMD search in conjunction with the long operational ranges present challenging timeline occupancy demands. Under some BMD scenarios, the radar may exhaust timeline for scheduling transmit and receive dwells before they reach their maximum long-term duty cycle. Hence, simulation tools must ‘‘keep book’’ of both duty cycle and timeline occupancy when estimating BMD radar resource usage. Figure 7-9 depicts the nominal cross-range resolution of selected BMD radars taking into account their estimated angular resolution (beamwidth) as well as typical operational ranges. In contrast to anticipated range resolutions of 1,000’s km Ground Clutter Range Range-Doppler Not to Scale

The exoatmospheric threat complex can be characterized as an ensemble of ballistic targets possibly accompanied by volumetric clutter. BMD radars are expected to resolve the threat complex in range and Doppler as depicted in Figure 7-10. The coarse grid over the threat complex corresponds to the notional resolution of surveillance radar, while the fine grid corresponds to that of an engagement sensor that must perform precision tracking and discrimination. The dark shapes correspond to discrete objects, while the shading denotes volumetric clutter that may be masking the discrete objects.

7.5.2 Track Prediction Performance Prediction accuracy is a critical performance metric for BMD sensors. Predicted impactpoint estimation is obviously critical to characterizing an attack as well as distinguishing between ballistic missiles and space-vehicle launchers. Prediction accuracy is critical to

314

CHAPTER

7

Ballistic Missile Defense Radar rs surement Erro ed Target Mea Uncompensat

FIGURE 7-11 ¢ BMD Tracking Process.

Smoothing ) Interval (Ts Track Smoothed Error with “Bias” Compensation

Predition Interval (Tp)

Measurement “Bias”

Track Prediction Error

Measurement Compensation

Track Filtering

External Coordinate Transformation

handover and system-level track correlation among the surveillance and engagement sensors that observe different portions of the threat trajectory. Designation of the target from these sensors to the interceptor to support fly-out trajectory planning as well as seeker acquisition by the kill vehicle at endgame requires prediction based on precision track data. Target tracking must also support a prediction interval adequate to compensate for latencies in the BMD-level command, control, and communications process. High-fidelity simulations anchored on test data are required for definitive performance assessment, but the expressions in this section provide some useful rules of thumb as well as analytical insight. This BMD tracking process is depicted in Figure 7-11. Returns are collected over the smoothing interval for subsequent measurement compensation and track filtering. At the end of the smoothing interval over which the target is tracked, its position is estimated from track filtering the sequence of measurements, resulting in sufficient reduction of the random error relative to that associated with a single measurement. The predicted target position is extrapolated over a specified prediction interval so that the resultant error is typically dominated by the velocity error at the end of the smoothing interval as delineated shortly. Measurement compensation is the correction process to remove bias-like errors, including both instrumentation errors due to the design of the radar as well as tropospheric refraction errors imposed by atmospheric propagation. True bias error would be spatially and temporally invariant such that it should be eliminated from a wellcalibrated radar system. More properly speaking, these are systematic residual errors that remain after nominal alignment and calibration but tend to be spatially or temporally correlated. Such errors tend to vary ‘‘slowly’’ relative to the track update rate and so manifest as bias-like error components that cannot be effectively smoothed by the track filtering process. These systematic residual errors can contribute to coordinate system misregistration and degrade system track performance. Modern error computation techniques supported by extensive reference target tracking and other calibration measures can reduce systematic residual error due to hardware sources to the same order as the nominal random error component. While the

7.5

BMD Radar Performance Estimation

Global Positioning System provides excellent positioning and timing data, it cannot directly characterize orientation error that manifests as angle error in the subject radar system. The sensitivity of BMD radars enables usage of satellites with precision ephemeris data as reference targets to characterize radar measurement error to support subsequent compensation. Frequent collection of calibration data also supports health and status reporting needs for BMD radar systems. Lookup table techniques that plot refraction bias versus elevation angle that are supplemented to account for diurnal and seasonal variations can eliminate roughly 90 percent of the mean tropospheric error. Computationally intensive model-based techniques anchored on local meteorological conditions can further suppress residual tropospheric error. The compensated measurements are processed by a track filter to suppress random measurement error due primarily to additive thermal noise. In addition, the filter is typically designed to estimate velocity to support track prediction. Track filter design must generally bound the degree of smoothing to accommodate unanticipated target maneuvers. However, exoatmospheric track filters designed for the ballistic missile coast phase benefit from the associated deterministic trajectory to achieve order-ofmagnitude improvement in smoothing performance over that feasible with a conventional track filter that must accommodate maneuvering targets. Evaluating tracking performance requires high-fidelity simulation tools anchored on test data to characterize the random and systematic residual accuracy of a given sensor. However, the following procedure can be used to coarsely assess track performance. As a point of departure, we will look at noise-limited performance (random error smoothing) as that is the fundamental bound remaining if alignment and calibration procedures are sufficient to suppress systematic residual and coordinate registration errors. Angle error effects dominate those of range error in measurement positions. Modern radars can readily measure range to a precision of meters or tens of meters. The effective cross-range measurement accuracy can be represented as the product of the azimuth and elevation angle errors in radians multiplied by the target range in meters. Neglecting such effects as radar line-of-sight rotation over the track interval [38] and detailed design parameters such as monopulse slope and antenna taper weighting [39], we can coarsely represent the standard deviation, sc, of the dominant one-dimensional cross-range error as lR sc rffiffiffiffiffiffiffiffiffiffi S D 2 N

(7-1)

where l ¼ wavelength, R ¼ range to the target, D ¼ antenna diameter, and S ¼ signal-to-noise ratio of the measurement referenced to the output of the signal N processor. Assuming that the angle measurement error is on the order of a tenth of a beamwidth and approximating beamwidths and operational ranges from Figure 7-9 suggests effective cross-range measure errors will be orders of magnitude larger than the range measurement error. Hence, the effective cross-range accuracy should dominate the composite track accuracy.

315

316

CHAPTER

7

Ballistic Missile Defense Radar

Assuming N measurements are taken at fixed intervals of T over the smoothing period Ts such that N > 1, the standard deviation of the smoothed cross-range at the end of the tracking interval can be approximated as 2sc scs pffiffiffiffi N

(7-2)

The corresponding standard deviation, scp, of the predicted cross-range for prediction interval Tp where Tp >> Ts can be estimated as pffiffiffiffiffi 12sc Tp scp pffiffiffiffi (7-3) N Ts Note that prediction error can be decreased by extending the smoothing interval even while keeping the number of measurements constant by increasing the measurement interval. While these expressions are based on theoretical bounds, they should be representative of the tracking performance of an optimal track filter matched to a target with a deterministic trajectory such as an exoatmospheric ballistic missile [40]. The composite two-dimensional cross-range error standard deviation can be estimated as the root-sum-square of cross-range standard deviation computed for azimuth and elevation. If the beamwidth is roughly symmetric, then the two-dimensional effective cross-range error can be estimated simply as the product of H2 and the onedimensional values given previously. This value can be appropriately inflated to account for systematic residual errors [41]. As an example of the utility of this calculation, the lateral delta velocity, DV, required by the KV to compensate for the radar prediction error, can be roughly estimated as [42] DV ¼

scp N 0 ðN 0 1ÞTf

(7-4)

where N0 ¼ Effective Navigation Ratio of the KC guidance system and Tf ¼ time-of-flight of the homing interval. This expression enables the radar prediction error to be traded against the homing interval and hence the effective range of the KV seeker. Assuming a lower bound of N0 ¼ 3 to ensure that endgame errors are closed out by the KV guidance process and a homing time of 10 s indicates that the composite radar prediction error must be kept to within ~330 m to keep the required DV requirement less than 50 m/s.

7.5.3 BMD CSO Performance As previously addressed, BMD radars are typically evaluated in terms of their noiselimited performance. This practice is in contrast to air-defense applications where performance is often determined by the radar capability to detect small targets in the presence of strong ground clutter. As described previously, the absence of drag in the exoatmospheric midcourse regime enables ‘‘fly along’’ volumetric clutter that may include debris as well as intentional masking countermeasures such as chaff. At long ranges, this ensemble of CSOs associated with a given ballistic missile launch will be contained within a single beamwidth of a BMD radar. The RV may be embedded in an

7.5

BMD Radar Performance Estimation

extended volumetric clutter field composed of these multiple sources. In addition to distributed clutter, multiple discrete targets, such as decoys, may be closely spaced with respect to each other and the RV, further challenging the track and discrimination processes. High-resolution range and tracking waveforms may be required to support tracking and discrimination as well as resilient algorithms. This section formulates performance measures to characterize CSO and volumetric clutter performance impacts. In air-defense applications, measures of surface clutter rejection indicate the radar’s capability to detect small, low-flying aircraft and cruise missiles. The clutter improvement factor (CIF), the ratio of signal-to-clutter ratio (SCR) at the signal processor output referenced to the SCR at the signal processor input for a given target range rate is commonly used to characterize this performance. CIF performance can be accurately modeled and substantiated with hardware measurements so it can be used to anchor anticipated radar-system performance. An implicit condition of CIF and similar performance measures is that target returns can be resolved in Doppler from the nearstationary surface clutter. CIF does not seem a general metric for evaluating BMD radar performance under volumetric clutter conditions since the clutter is not necessarily Doppler resolved from targets of interest. The total integrated RCS of such volumetric clutter is likely orders of magnitude less than near-in surface clutter so CIF requirements do not drive RF dynamic range in BMD as they do in air defense. CIF estimation could be used to characterize the signalprocessing benefits of ‘‘thinning out’’ clutter thru extended coherent integration under a given scenario. In BMD applications, the measurements must be compensated for the estimated target velocity to maintain the target in a given range gate to enable pulse integration. Objects with dissimilar range-rates will ‘‘fly through’’ multiple range gates over the integration interval so that their returns are correspondingly ‘‘smeared.’’ This condition occurs whenever the product of the object’s range-rate offset from the target under track and the pulse integration interval exceeds the range-gate extent, which should be on the order of the radar-range resolution. The challenge of volumetric clutter is its proximity in range and Doppler to targets of interest, which challenges resolution and measurement-to-track data association. In contrast to surface clutter, the range and Doppler distribution of BMD volumetric clutter cannot be delineated a priori since the birthing and distribution are determined by the design and operation of the ballistic missile threat. Moreover, volumetric clutter may vary dynamically over the engagement interval due to discrete events such as booster breakup or chaff release. The range or range-rate spread in the dispersion of the threat complex may include ‘‘crossing targets’’ where the radar must maintain track on individual CSOs that overtake each other in range, resulting in merged measurements and misassignment of measurements to tracks. Resolution under crossing target conditions is parameterized in Figure 7-12. These curves compare performance achievable with range and Doppler processing in the case of a given target overtaking another as a function of differential range rate. The depicted time interval is defined as 1. the period over which the two objects are obscured (unresolved) in range for a radar employing a given waveform bandwidth, or 2. the required coherent-processing interval required to achieve continual Doppler resolution of the objects.

317

FIGURE 7-12 ¢ Obscuration Interval and Coherent Processing Interval for Crossing Targets.

CHAPTER

7

Ballistic Missile Defense Radar

10,000.0 1,000.0 Time (ms)

318

Obscuration Duration (Unresolved in Range) for Measurement Bandwidths {10, 100, 1,000} MHz

100.0 10.0 1.0

Minimum X-band CPI Required for Doppler Resolution

0.1 0

20

40 60 80 Relative Range Rate (m/s)

100

In both cases, the resolution was calculated as twice the Rayleigh limit to account for broadening factors such as tapering to suppress range sidelobe or Doppler sidebands, RF band limiting, and signal-processing artifacts. Hence, the following pragmatic relationships will be used: Pragmatic range resolution ¼ 2C=ð2W Þ ¼ C=W Pragmatic Doppler resolution ¼ 2=Tci where W ¼ waveform bandwidth and Tci ¼ coherent pulse integration interval. As depicted, coherent processing across a pulse train to achieve Doppler resolution of a pair of targets with a given range-rate differential reduces the obscuration interval relative to what can be achieve with single-pulse range-only processing. This result is somewhat intuitive as Doppler resolution derives from the time required for the differential range to change on the order of a wavelength while the range-only processing for a given obscuration interval is predicated on a relative motion of tens or hundreds of wavelengths. Moreover, the coherent processing maintains continual Doppler resolution of the crossing targets; tracking on range-only measurement processing mandates extrapolating or ‘‘coasting’’ the tracks through the obscuration interval until the composite targets can be resolved again. Tracking performance in volumetric clutter is highly dependent on the clutter distribution and density as well as the radar waveforms, signal processing, and data-association implementation. Data association is the process of assigning M measurements from a given dwell to update N tracks, where M and N may differ due to missed detections, false alarms, clutter detection, and track-management artifacts. The trackmanagement process may inadvertently create redundant or merged tracks as well introduce spurious track artifacts though track initiation and deletion operations [43]. A number of algorithmic approaches to data association have been suggested across a wide span of effectiveness and computational demands. At a fundamental level, tracking requires well-resolved measurements that are correctly associated. The composite probability of both resolving and correctly associating measurements is termed purity. Anticipated purity can be estimated from target

7.5

BMD Radar Performance Estimation

319

density, measurement characteristics, and tracking accuracy. While actual performance verification must be anchored using high-fidelity simulation tools, a number of analytical expressions are useful for coarse performance estimation [44]. Conventional tracking algorithms need a composite purity of ~0.9 to maintain firm tracks. Estimating purity in a BMD context can be useful for coarse performance assessments. Exoatmospheric track filter design can exploit the deterministic knowledge of a ballistic trajectory to achieve a much higher degree of smoothing than is achievable in air-defense applications where target maneuvers must be accommodated. The position of a ballistic target can be predicted to the next update measurement opportunity to well within a range gate. The probability of correct measurement association is correspondingly enhanced since the track-update acceptance gates for measurements decrease as the track prediction error decreases, which reduces the probability of misassociation in dense target environments. Data-association performance is further improved when tracking adjacent objects as well as specific targets of interest. Blackman has shown that a given level of dataassociation performance can be maintained at H2 to 4 times the tracked target density relative to a target embedded in an equivalent false alarm density, depending on the measurement dimensionality [45]. However, this requirement can result in significant computational loading on the data processor Mutliple object. As an illustrative analysis metric, we shall exploit the simplified scenario depicted in Figure 7-13 of a notional upper-stage fragmentation that produces No objects randomly distributed across a spherical surface expanding at a constant rate and producing multiple crossing targets with a warhead under track. Assuming that the fragments are all contained within the radar beam, the mean number of objects in a given range-gate extent equivalent to the range resolution, nr, is given simply by nr ¼

No dr 2r

(7-5)

where dr ¼ range resolution and r ¼ radius of the fragment sphere, which can simply be parameterized as the product of the expansion velocity and time after fragmentation. The probability of resolution is defined here as the probability of a given range gate containing 0 objects such that a target under track at that range would be

Wideband Resolution Narrowband Resolution

FIGURE 7-13 ¢ Spherical Fragmentation Example for Evaluating Purity.

320

CHAPTER

7

Ballistic Missile Defense Radar

unimpaired. The number of objects in a given range gate is represented by a Poisson probability density function so that the probability of resolution, Pres, can be calculated as [46] Pres ¼ Pfx ¼ 0g ¼ enr

nxr ¼ enr x!

(7-6)

Under the same scenario, the probability of correctly data association, Pca, can be derived as Pca ¼ e

pnr p

(7-7)

where e ¼ ratio of the range-gate extent to the standard deviation of the estimated range predicted at the time the measurement was taken. This expression corresponds to the mean number of objects within a one-sigma prediction-update volume for the one-dimensional case applicable to this example. As depicted in Figure 7-14, the purity of a mature track is dominated by resolution performance rather than data association for established tracks, even assuming conservative track filter smoothing with e ¼ 4. The composite purity appears inadequate to maintain track when the mean number of objects in the range gate exceeds ~0.1, corresponding to the objects spaced in range at ten times the range resolution. Canonical analysis shows that probability of resolution should be consistently lower than the probability of correct association under BMD tracking conditions [47]. Resolution is also a prerequisite to initiating tracks in a stressing CSO environment. Both resolution and data-association performance can be improved by decreasing dr, corresponding to increasing tracking waveform bandwidth.

1.0 0.9 0.8 0.7 Probability

FIGURE 7-14 ¢ Probability of Resolution and Probability of Correct Association for Spherical Fragmentation Example.

0.6 0.5 0.4 0.3 0.2 0.1 0.0 0.01

0.10

1.00

Mean Number of Objects / Range Resolution

10.00

7.6

7.6

References

REFERENCES

[1] J. Constant, Introduction to Defense Radar Systems Engineering, Spartan Books, New York Washington 1972. [2] D. Ausherman, A. Kozma, J. Walker, H. Jones, and E. Poggio, ‘‘Developments in Radar Imaging,’’ IEEE Transactions AES, 20, no 4, July 1984. [3] J. Spencer, The Ballistic Missile Threat Handbook, Heritage Foundation, Washington, 2000. [4] R. Braham, ‘‘Ballistic Missile Defense: It’s Back,’’ Spectrum, Vol 31, No 9, September 1997. [5] Spencer, ibid. [6] P. Mantle, The Missile Defense Equation: Factors for Decision Making, AIAA, Reston VA, 2004. [7] P. Zarchan, ‘‘Ballistic Missile Defense Guidance and Control Issues,’’ Space & Global Security, 1998, 8, pp. 99–124. [8] B. Graham, Hit to Kill: The New Battle over Shielding America from Missile Attack. Public Affairs, New York, 2001. [9] Y. Velikhov et al., Weaponry in Space: The Dilemma of Security. MIR Publishers, Moscow, 1986. [10] T. Lin, ‘‘Development of U.S. Air Force Intercontinental Ballistic Missile Weapon Systems,’’ Journal of Spacecraft and Rockets, 40, no. 4, July–August 2003. [11] G. Lewis and T. Postol, ‘‘Future Challenges to Ballistic Missile Defense,’’ Spectrum, Vol 34, No 9, September 1997. [12] Sessler et al., ‘‘Countermeasures: A Technical Evaluation of the Operational Effectiveness of the Planned U.S. National Missile Defense System,’’ Union of Concerned Scientists, M.I.T., April 2000. [13] M. Sarcione et al., ‘‘The Design, Development and Testing of the THAAD (Theater High Altitude Area Defense) Solid State Phased Array,’’ IEEE Symposium on Phased Array Systems and Technology, Boston, 1996. [14] J. Tomay, ‘‘Warning and Assessment Sensors,’’ Managing Nuclear Operations, A. Carter et al., editors, Brookings Institution, 1987. [15] ‘‘Visit to Fylingdale Moor BMEWS,’’ http://www.subbrit.org.uk/rsg/sites/f/fylingdales/. [16] E. Filer and J. Hartt, ‘‘Cobra Dane Wideband Pulse Compression System,’’ EASCON, 1976. [17] M. Sarcione, ibid. [18] S. Boyev, ‘‘Shrewd Eyes and Mind of the Missile and Space Defense High-Potential Radars: Past, Present and Future,’’ Military Parade, September 2001. [19] P. Podvig, editor, Russian Strategic Nuclear Forces, M.I.T. Press, Cambridge, MA, 2001. [20] P. Podvig, ‘‘History and the Current Status of the Russian Early-Warning System,’’ Science and Global Security, 10, pp. 21–60, 2002. [21] Soviet Military Power 1988, U.S. Government Printing Office. [22] Podvig, ibid. [23] Podvig, ibid. [24] Podvig, ibid. [25] P. Podvig, editor, Russian Strategic Nuclear Forces, M.I.T. Press, Cambridge, MA, 2001. [26] http://www.globalsecurity.org/wmd/world/russia/pill_box.htm.

321

322

CHAPTER

Ballistic Missile Defense Radar

7

[27] Boyev, ibid. [28] Podvig, ibid. [29] B. Naveh and A. Lorber, editors, Theater Ballistic Missile Defense, AIAA, 2001. [30] S. Glasstone and P. Dolan, The Effects of Nuclear Weapons, U.S. Government Printing Office, 1977. [31] Constant, ibid. [32] Radiation Intensity of the PAVE PAWS Radar System, National Academy of Science, 1979. [33] B.-Z. Naveh and A. Lorber, editors, Theater Ballistic Missile Defense, AIAA, Reston, VA, 2001. [34] Y. Bar-Shalom and X.-R. Li, Mutlitarget-Multisensor Tracking: Principles and Techniques, YBS, 1995. [35] A.H. Register, W.D. Blair, and G. Brown, ‘‘Improved Radar Revisit Time Control to Observe Ballistic Missile Mode Transitions,’’ Proceedings of the Thirty-Fourth Southeastern Symposium on System Theory, 2002. [36] C. Wilson, High Altitude Electromagnetic Pulse (HEMP) and High Power Microwave (HPM) Devices: Threat Assessments, Congressional Research Service August 20, 2004. [37] R. Ghose, EMP Environment and System Hardness Design, Interference Control Technologies, Gainesville, VA, 1984. [38] F. Daum and R. Fitzgerald, ‘‘Decoupled Kalman Filters for Phased Array Radar Tracking,’’ IEEE Transactions on Automatic Control, 28, no. 3, March 1983. [39] D. Barton and H. Ward, Handbook of Radar Measurement, Artech, Dedham MA, 1984. [40] J. Minkoff, Signals, Noise, and Active Sensors: Radar, Sonar, Laser Radar, Wiley, 1992. [41] D. Hsu, Spatial Error Analysis: A Unified Application-Oriented Treatment, IEEE Press, New York, 1998. [42] P. Zarchan, Tactical and Strategic Missile Guidance, Third Edition, AIAA, Reston, VA, 1997, p. 292. [43] K. Manson and P. O’Kane, ‘‘Taxonomic Performance Evaluation for Multitarget Tracking Systems,’’ IEEE Transactions, 28, no. 3, July 1992. [44] F. Daum, ‘‘A System Approach to Multiple Target Tracking,’’ Y. Bar-Shalom, editor, Multitarget-Multisensor Tracking: Applications and Advances, Volume II, Artech, Dedham MA, 1992. [45] S. Blackman and R. Popoli, Design and Analysis of Modern Tracking Systems, Artech, Dedham MA, 1999, p. 885. [46] M. Belcher, ‘‘Tracking Unresolved Targets in Theater Ballistic Missile Defense,’’ Proceedings of the 29th Southeastern Symposium on System Theory, March 1997. [47] Blackman and Popoli, ibid.

7.7

FURTHER READING

S. Weiner and S. Rocklin, ‘‘Discrimination Performance Requirements for Ballistic Missile Defense,’’ Lincoln Laboratory Journal, 7, no. 1, 1994. T. Jeffrey, Phased-Array Radar Design: Application of Radar Fundamentals, SciTech Publishing, Raleigh, NC, 2009.

CHAPTER

Ground-Based Early Warning Radar (GBEWR): Technology and Signal Processing Algorithms

8

A. De Maio, Universita` degli Studi di Napoli ‘‘Federico II’’, Italy, A. Farina and L. Timmoneri, Selex ES - A Finmeccanica Company, Italy and M. Wicks, University of Dayton (OH), US

Chapter Outline 8.1 8.2 8.3 8.4 8.5 8.6 8.7 8.8 8.9

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Phased Array Antenna . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Transceiver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Waveforms and Signal Processing . . . . . . . . . . . . . . . . . . . . . . . Tracking. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Electronic Counter-Countermeasures (ECCM) Capabilities . . . . . Special Functions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions and Further Reading. . . . . . . . . . . . . . . . . . . . . . . . References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

8.1

........ ........ ........ ........ ........ ........ ........ ........ ........

.... .... .... .... .... .... .... .... ....

323 335 342 348 352 357 359 376 377

INTRODUCTION

Radar systems can be classified into two main families according to their application: defense and civilian. Defense radars include air-defense and battlefield radars, whereas civilian radars include weather radars, radars for remote sensing, and others. Air traffic control (ATC) radars can be either defense or civilian. Air defense radars are also subdivided into two more categories: two-dimensional (2-D) and three-dimensional (3-D). The former can measure the target range and azimuth. The latter can determine the target’s height as well. Since air-defense radars can detect targets at great distances (hundreds of kilometers), they can be used as early warning radars (EWRs). A classification of EWRs can be provided based on the platform where the radar is installed. Then, it is possible to distinguish between airborne and ground-based (GB) EWRs. In this chapter, GBEWRs will be discussed. A GBEWR is used primarily for the long-range detection of targets, i.e., allowing defenses to be alerted as early as possible before the intruder reaches its objective, giving 323

324

CHAPTER

8

Ground-Based Early Warning Radar (GBEWR)

the defenses the maximum time in which to operate [1]. The most challenging targets for a GBEWR are ballistic targets (BTs) because of their small radar cross section (RCS) and the great ranges (hundreds and sometimes even thousands of kilometers) where they are required to be detected. The GBEWRs have a low-measurement resolution, work with lowpulse repetition frequency, and use low-carrier frequencies. The choice of low frequency is supported by the low propagation loss and by the high power that can be transmitted. On the other hand, the resolution and accuracy are limited because a lower frequency requires very large physical antennas. In particular, they work at L- and S-bands (1–2 GHz/2–4 GHz) [2]. The best advantage for choosing these bands over very high frequency (VHF) is related to the smaller size of antenna that allows a better compromise between performance and deployability. In addition, the GBEWR can guarantee timely and reliable detection of a high number of targets, which are generally embedded in heavy natural (clutter) and manmade (electronic countermeasures (ECM)) interferences. The target RCS can range from very low to relatively high figures. Discriminating false detections from true targets and high accuracy in position finding, even in a multitarget environment (e.g., mass raid), are additional requirements. Finally, target classification, threat assessment, and efficient dissemination of processed data to computer and display systems are facilities that shall be implemented. This chapter is organized as follows. Section 8.1 introduces the GBEWR and its main characteristics. Section 8.2 introduces the phased array antenna and illustrates its main characteristics, with a special focus on some beam-scanning techniques. Classic and modern architectures of the transceiver and its main parameters are presented in Section 8.3. Section 8.4 provides an overview of several signal-processing algorithms used in GBEWRs. Section 8.6 describes the issue of tracking, with emphasis on the evolution of both filtering and correlation logics. Section 8.7 provides a description of the ECCM capabilities and, in particular, of the sidelobe blanking technique. In Section 8.8, several GBEWRs’ special functions are reported. Specifically, this section deals with the problem of radar detection and tracking of a BT, of the low probability of intercept, and of the denial of bistatic hosting by waveform design. Finally, conclusions are drawn in Section 8.9. The following is a list of abbreviations and acronyms used in this chapter.

8.1.1 Acronyms/Abbreviations 2-D 3-D ABT ACCS ACE ADC AEGIS AFS AGARD MSP ALTBMD AMS AMTI

two-dimensional three-dimensional air-breathing target Air Command and Control System Allied Command Europe analog-to-digital converters Airborne Early Warning Ground Environment Integration Segment automatic frequency selection Advisory Group for Aerospace Research & Development Mission System Panel active layered theater ballistic missile defense Alenia Marconi Systems adaptive moving target indicator

8.1

ARM ATC BFN BITE BM BMD BMEWS BST BT CA-CFAR CAD CFAR COHO CUT CW C3 DAC dB DDC DOA DUC ECM ECCM EKF ESM EuRAD EW EWR FA FFT FPGA GaAs GBEWR GIS GO-CFAR HF HMI HPRF Hz ICBM IEE

anti-radiation missile air traffic control beam-forming network built-in test equipment ballistic missile ballistic missile defense ballistic missile early warning system beam-scanning technique ballistic target cell averaging constant false alarm rate computer-aided design constant false-alarm rate coherent oscillator cell under test continuous waveform Command and Control Center digital-to-analog converter decibel digital down conversion direction of arrival digital up conversion electronic countermeasures electronic counter-countermeasures extended Kalman filter electronic support measures European Radar Conference early warning early warning radar false alarm fast Fourier transform field programmable gate array gallium arsenide ground-based early warning radar geographic information system greatest of constant false-alarm rate high frequency human machine interface high pulse repetition frequency Hertz intercontinental ballistic missile Institution of Electrical Engineers

Introduction

325

326

CHAPTER

IEEE IET IF IFF IFFT ILDC IMM ISL ITRS JPDA KB KF LFM LNA LO LP LPI LPRF MESFET MHT MHz MIS MM MMIC MMSE MTI/MTD MW NADGE NAEW NAI NF NLFM NN ONS OS-CFAR OTHB OTHR PAAM PDA PIN-diode PO PRF

8

Ground-Based Early Warning Radar (GBEWR) Institute of Electrical and Electronics Engineers Institution of Engineering and Technology intermediate frequency identification friend or foe inverse fast Fourier transform incremental length diffraction coefficients interactive multiple model integrated sidelobe level International Technology Roadmap for Semiconductors joint probabilistic data association knowledge-based Kalman filter linear frequency modulation low-noise amplifier local oscillator linear programming low probability of intercept low-pulse repetition frequency metal semiconductor field effect transistor multiple hypothesis tracking MegaHertz multiple independently scanned multiple model monolithic microwave integrated circuit minimum mean square error moving target indicator/moving target detector moving window NATO Air Defense Ground Environment NATO Early Warning not automatic initialization noise figure nonlinear frequency modulation nearest neighbor original NADGE sites ordered statistic constant false alarm rate over-the-horizon backscatter over-the-horizon radar phase and amplitude adjustment modules probabilistic data association P-type/intrinsic/N-type (doped semiconductor diode) physical optics pulse repetition frequency

8.1

PRT PSL PTD RAN RAT RCS RES RF RX SoC SFDR SLB SLBM SNR SO-CFAR SOCP SR SSC SSR STALO STC TBM ToT TR TRM TX UHF/VHF VS IMM WWII

Introduction

pulse repetition time peak sidelobe level physical theory of diffraction radar for naval surveillance radar for terrestrial surveillance radar cross section radar environment simulator radio frequency receiver system on a chip spurious free dynamic range sidelobe blanking sea-launched ballistic missile signal-to-noise power ratio smallest of constant false alarm rate second order cone programming shift register scan to scan correlation secondary surveillance radar stable local oscillator sensitivity time control tactical ballistic missile time on target transmit/receive transmit/receive module transmitter ultrahigh frequency/very high frequency variable structure interactive multiple model Second World War

8.1.2 Historical Perspective The first GBEWR was the British Chain Home, used during the Second World War (WWII) for the detection of the enemy’s aircraft [2, 3]. It was composed of radars on top of towers to provide long-range detection. This system was designed by Sir Robert Watson-Watt and was established by 1939. The Chain Home worked in a range of frequencies between 20 and 55 MHz, and had a maximum detection range of 190 km. Because the British Chain Home could not detect low-flying aircraft, the United Kingdom created a second system of GBEWRs called Chain Home Low, with an operating frequency of 180–210 MHz and a maximum detection range of 160 km. As a result, the two GBEWR chains provided a good coverage of the space. In the same years, other important GBEWRs were developed: U.S. SCR-270 (Signal Corps Radio model 270), U.S. AN/CPS-6, U.S. CXAM, and German FREYA [4]. All these systems had a maximum detection range smaller than 190 km because of the propagation loss and the

327

328

CHAPTER

8

Ground-Based Early Warning Radar (GBEWR)

limitation of the line of sight. The U.S. SCR-270 was the first long-range radar of the United States and was used for the first time in 1941 during the Pearl Harbor attack [5]. This EWR worked at an operating frequency around 100 MHz and had a maximum detection range from 120 to 190 km. There were two typologies of this radar: the mobile SCR-270 and the fixed SCR-271. Also the U.S. AN/CPS-6 was developed during WWII in 1945. Initially, the radar was designed to detect fighter aircraft for about 160 km; this radar operated at S-band frequencies from 2.7 to 3 GHz. The shipborne U.S. CXAM was used in 1941 in Australia and was positioned at the top of the ship’s mast. It was able to detect single aircraft at 80 km and to detect large ships at 22 km. The German FREYA (a semi-mobile system), whose operating frequency was 120–130 MHz, had a maximum detection range of 160 km and could not determine the altitude. Japan developed its own systems, Mark-1 and Tachi-6, with detection ranges of 120 and 200 km, respectively. In 1942, the USSR installed RUS-2 radars with a maximum detection range between 95 and 145 km to aid the local defense of Moscow and Leningrad. The use of GBEWRs grew during the Cold War. The United States and Canada in 1951 began the construction of more than 30 stations situated along their common border. Afterward, the Distant Early Warning Line (DEW Line), a series of radar stations, were installed along the Arctic Circle stretching from Alaska to Greenland. In 1985, the DEW Line was upgraded with new radars and was called the North Warning System. In the 2000s the U.S. Air Force upgraded operating software for long-range atmospheric GBEWR systems. During the 1970s, the U.S. Air Force had contracted with General Electric Aerospace to build a prototype over-the-horizon backscatter (OTHB) radar in Maine (AN/FPS-118). To counter the threat of surprise intercontinental ballistic missile (ICBM) attacks by the USSR, the United States also began the construction of the ballistic missile early warning system (BMEWS). Later, the United States replaced the BMEWS with the PAVE PAWS (developed by Raytheon Corp.), a long-range, solid-state phased array radar system [6] with the goal of assuring protection against sealaunched ballistic missile (SLBM) attacks. Meanwhile, the USSR developed several EW and OTH radar (OTHR) systems against BM attacks and for air defense and developed systems for EW of cruise missile and bomber attacks. For BM detection, the USSR developed 11 large HEN HOUSE BM EWRs that were deployed to six locations along the Soviet borders. With the USSR’s collapse in 1990, the missile warning radar system became too fiscally burdensome. In 2001, Russia announced its intention to scrap many of the fixed installations and rely on mobile stations. In Australia, OTHRs were developed starting in 1950 [7]. The Jindalee radar network was the first Australian skywave OTHR system. In addition, several radar sets were produced to suit the particular operational and environmentally extreme conditions of the Pacific Theater during WWII. For example, 1,300 units of the British-designed ASV MkII were produced [8]. Indeed, the EW’s capability of an OTH radar may be utilized to cue higher precision but more circumscribed surveillance and reconnaissance assets mounted on mobile platforms about areas where unauthorized activity has been detected. The complementary nature of such systems leads to more efficient use of defense resources for an effective response [9]. A series of NATO common-funded systems, called NADGE (NATO Air Defense Ground Environment) chain, are deployed in Europe [10].

8.1.2.1 The Evolution of NADGE The NATO Air Defense Ground Environment (NADGE) was introduced in the late 1960s to provide a modernized, semi-automated air defense system, comprising new radars, new

8.1

Introduction

ground-to-air communications, and computer-based control sites, with the ability to automatically exchange data between all the sites throughout Allied Command Europe (ACE). The software for the NADGE system was hosted on dedicated military computers. Over the past 30 years, the various systems that comprise the overall NADGE system have been updated to account for both the changing air threat and more capable sensors and weapon systems. A major upgrade took place when the NATO Airborne Early Warning (NAEW) aircrafts were introduced. The NADGE system was enhanced with the airborne early warning ground environment integration segment (AEGIS), thereby providing interoperability with the NAEW aircraft. An evolution of NADGE is represented by the Air Command and Control System (ACCS), which provides an in-place air command and control system scaled to accept air reinforcements and a deployable capability to enhance and augment the in-place facilities. ACCS provides services common to all sites supporting functions such as communications, information handling, data distribution, system management, interfaces, etc. Some innovative aspects in ACCS surveillance with respect to old NADGE capabilities include: ● ●

● ●

Multisensor tracker: All detections from all internetworked sensors are accounted for. One unique active/passive tracking (versus an active tracking and a concurrent passive tracking): Military and civilian radars, passive and active sensors, all feed the same tracker. High update rate. Online debiasing: Sensors and data-links are compensated for biases, so there is no need for manual intervention and calibration campaigns. Automatic identification data combining process: ACCS is the first system implementing this tool that will dramatically support air picture identification and track classification.

As a final remark, we mention that in 1998, NATO agreed to a program plan for an active layered theater ballistic missile defense (ALTBMD) capability to satisfy the military operational requirements. The objective is to defend NATO forces, deployed either within or beyond NATO’s area of responsibility, against the threat posed by tactical BMs, with ranges up to 3,000 km. At the NATO Summit 2010 in Lisbon (http://en.wikipedia.org/wiki/2010_Lisbon_summit), heads of state and government approved the alliance’s strategic concept and agreed to develop a BM defense capability with the aim of providing full coverage and protection for all NATO European populations, territories, and forces against the increasing threats posed by the proliferation of BMs.

8.1.3 Typical Characteristics GBEWR coverage is typically obtained by scanning in elevation up to 20 or 30 , depending on the application, while mechanically rotating in azimuth. The antenna has a quite low rotational speed,1 providing 360 azimuth coverage. Multiple independent and simultaneous pencil beams are often used in transmission and in reception. Accuracy of elevation angle measurement is guaranteed by monopulse technique. The receiver channel is linear in a wide

1

That is, 6 or 12 rounds per minute (rpm).

329

330

CHAPTER

8

Ground-Based Early Warning Radar (GBEWR)

range, and it has a high dynamic on the order of 80 dB. (see Section 8.3). It uses digital pulse compression techniques to ensure a suitable average power with low-peak power to provide resistance against anti-radiation missile (ARMs) without a reduction in range resolution. The sensitivity time control (STC) avoids signal saturation when reflection is too high. The aforesaid features, when added to frequency agility, allow the radar to operate in an intense clutter environment and ECM. The ECCMs are assured by low sidelobe antenna, sidelobe blanking (SLB), the availability of a large number of different frequencies, frequency analysis on azimuth sector, automatic frequency selection (AFS), random selection, and jam strobe reporting. Very low antenna sidelobes attenuate the interference outside the mainlobe. Fixed and adaptive moving target indicator/moving target detector (MTI/MTD) filters allow the operator to update suitable clutter maps, optimizing radar performances against the worst environmental conditions. The filters are generally enabled by continuously updated maps to optimize the performance for ground and sea clutter, rain, chaff, and clear conditions. The tracker manages several hundred 3-D tracks with the assistance during the plotgeneration process of the false-alarm control function. EWR needs be easily integrated in all combat management systems, and it has generally to be completely remote controlled.

8.1.4 Platform Usually, a radar classification on the base of the hosting platform is given. Thus, there are ground-based, shipborne, airborne, and spaceborne radars. In the next sections, ground-based radars are presented.

8.1.4.1 GBEWR: Some Commercial Systems Ground-based radars do not suffer the problem of power limitation that shipborne or airborne radars do this feature enables them to cover large areas of space [11, 12]. Ground-based radars are characterized by a high antenna gain, which provides high incident power on the target, and a high effective antenna receiving area. The principal characteristics of a typical GBEWRs are presented in Section 8.1.6. Actually, there are several commercial GBEWRs, both fixed and mobile. The following is a list of the principal commercial fixed ground-based long-range radars: – Selex EX RAT 31 DL [13] is an advanced L-band, solid-state, phased array, 3-D surveillance radar. Figure 8-1 shows a picture of the system antenna. Detection characteristics: ●

Instrumented range: 500 km

Elevation coverage: 20

Beam-scanning techniques: Multiple simultaneous independently phase-controlled pencil beams. Each beam provides monopulse altitude measurements with excellent accuracy, even in the frequency agility mode. Clutter suppression: The system presents an anti-clutter filter with adaptive notch MTI cancellers, fixed and real-time automatic clutter maps, with a moving window azimuth correlator and a digital modulus extractor for the amplitude detectors. ECCM: Reduced peak power provides resistance against ARM and ECM. Excellent ECCMs are provided by very low sidelobe antenna, sidelobe blanking,

8.1

Introduction

331 FIGURE 8-12 ¢ RAT 31 DL L-band, Solid-State, Phased Array, 3-D Air Surveillance Radar [Courtesy of Selex ES].

reduced peak power, frequency agility, jam strobe reporting, and a separate receiver for ECM monitoring. Other characteristics: The RAT 31 DL has 42 transmitter/receiver modules and several simultaneous independent pencil beams in elevation, and a 5/6 rpm azimuth scan rate. It is also fully transportable. The receiver is a double-conversion, superetherodine, dual-matched model, with a frequency selection that can be manual, random, or automatic. – THALES RAYTHEON SYSTEMS GROUND MASTER 400 [14] is an S-band, fully digital, long-range, radar. – LOCKHEED MARTIN FPS-117 [15] is an L-band, long-range, solid-state radar for both air surveillance and en route air traffic control. – BAE SYSTEMS S743D MARTELLO [16] is an L-band, long-range, transportable/fixed-site, 3-D air defense surveillance radar. – INDRA 3D LANZA [17] is an L-band, multiscenario, multithreat adaptive radar. – BAE SYSTEMS COMMANDER SL [18] is an S-band, long-range, tactical air defense radar.

8.1.4.2 Mobile GBEWR: Some Commercial Systems Here, an overview of the main commercial mobile ground-based long range radars is presented. – Selex ES RAT 31DL/M [19] is an L-band, 3-D, fully solid-state, tactical longrange EWR and a tactical BM defense radar. Detection characteristics: ●

Elevation coverage: from –2 to 20

Instrumented range: 400 km

Azimuth coverage: 360

2 This figure and those from 8-3 to 8-16 are either from references [23]–[26] and from the website http://www.selex-es.com/

332

CHAPTER

8

Ground-Based Early Warning Radar (GBEWR)

Beam-scanning techniques: The radar architecture exploits multiple simultaneous pencil beams (MSPB). The multiple beams are electronically and independently steered in elevation, both in transmission and in reception. Monopulse technique is adopted for the elevation angle measurements. Clutter suppression: The MSPB architecture provides a large number of transmitted pulses in each beam-pointing direction, guaranteeing high clutter suppression in adverse weather conditions in the whole instrumental coverage volume. ECCM: The most advanced processing techniques are supported by flexible and state-of-the-art signal and data processors. These features, combined with the ultralow antenna sidelobes, guarantee an outstanding jamming resistance. The MSPB technology allows innovative and dedicated war-fighting solutions against several threat sources such as tactical ballistic missiles (TBMs), ECMs, and mass raids. Other characteristics: High mobility; transportability: aircraft (C-130), helicopter (CH-47), road (10 tons std); the radar is equipped with its own electrical power source and is self-sufficient for a long time. The system acts as an air defense stand-alone Command and Control Center (C3). It can be integrated in a cluster of some netted RAT 31 family systems reporting to a mobile C3, which ensures outstanding radar cooperation by means of robust radio link communications. – THALES RAYTHEON SYSTEM MASTER-M [20] is an S-band, long-range, infrastructure 3-D surveillance radar. – LOCKHEED MARTIN AN/TPS-77 [21] is an L-band, phased array, all-solidstate radar. – BAE SYSTEMS AR327 COMMANDER [16] is an S-band, land-based mobile air defense radar.

8.1.4.3 Deployable GBEWR Deployable GBEWRs are commonly referred to as tactical long-range radar, operating in the L- or S-band, specifically designed to support peacekeeping missions. They shall be deployed on the battlefield as a front-line system to protect and survey territories and assets against all air threats. To perform these tasks in a worldwide tactical environment, the deployable GBEWR shall be highly mobile and shall not require any special loading/ unloading equipment, i.e., the radar is deployable with its own built-in means and tools, and including its self-installing devices not involving cranes or other tools. The whole system has to be housed in a few (say, two) 20-ft ISO containers mounted on two commercial cross-country trucks for land mobility. They are designed to provide users with the capability to move quickly for regrouping and are generally equipped with their own electrical power source and are autonomous for a long time. These systems act as a stand-alone C3. They must be integrated in a cluster of other systems reporting to a mobile C3, which ensures outstanding radars cooperation by means of robust radio link communications.

8.1.5 Requirements for GBEWR Table 8-1 contains the main performance requirements of a fixed GBEWR. In addition to the detection probability, the track initiation probability (Pti) is of paramount importance for GBEWR. Pti represents the probability that a firm track is

8.1 TABLE 8-1

¢

Introduction

Requirements for fixed GBEWR

Coverage volume

Typical coverage volume is 360 in azimuth 20 in elevation 450 km instrumented range

Detection range

Assuming free space propagation, the typical required detection range of the primary radar is 300 km.

Measurement accuracies

The typical three-dimensional measurement accuracies of the primary radar are in range: 50 m in azimuth: 0.3 in height: 400 m

Measurement resolutions

The typical three-dimensional measurement resolutions of the primary radar are in range: 300 m in azimuth: 3 in height: 3

Clutter cancellation

Typical requirement is 50 dB for ground clutter.

1

FIGURE 8-2 ¢ Detection Probability and Track Initiation Probability Versus Target Range for a Typical GBEWR.

Detection Probability

Detection and Track Initiation Probabilities

0.9

Track Initiation Probability

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 200

333

250

300 Range (km)

350

400

established for a detected target at a certain distance. The Pti is estimated starting from the detection probability, having fixed a certain N out of M logic, i.e., N independent detections should happen in a batch of M possibilities. The N out of M value defines the probability of false track, which is generally in the order of one per hour. Let us consider a typical GBEWR and a 1 m2 RCS target. Figure 8-2 reports the detection probability of the radar (blue curve) and the track initiation probability (red curve) when a 3 out of 4 logic is applied. Common requirements call for a Pti of 90 percent, which is achieved at a distance of about 290 km.

334

Ground-Based Early Warning Radar (GBEWR)

CHAPTER

8

TABLE 8-2

Requirements for mobile GBEWR

¢

Coverage volume

Typical coverage volume is: 360 in azimuth 30 in elevation 400 km instrumented range

Detection range

Assuming free space propagation, the typical required detection range of the primary radar is 250 km.

Measurement accuracies

The typical three-dimensional measurement accuracies of the primary radar are: in range: 50 m in azimuth: 0.4 in height: 400 m

Measurement resolutions

The typical three-dimensional measurement resolutions of the primary radar are: in range: 300 m in azimuth: 4 in height: 4

Clutter cancellation

The typical requirement is 50 dB for ground clutter.

To achieve its mission and reach the performance requirements of Table 8-1, a GBEWR shall possess a number of unique characteristics, which will be discussed in the next sections. Table 8-2 contains the main performance requirements that a mobile GBEWR should possess.

8.1.6 Distinctive Characteristics of a GBEWR The GBEWR is typically a fully solid-state radar with transmitter modules distributed along the antenna array. The radar system is a self-contained unit of modular design to facilitate damage repair by rapid replacement of modular subsystems and subassemblies. The design of the radar shall allow both colocated and non-colocated installations with monitoring and control facilities. The expected key features of GBEWR include: – Fully solid-state radar with antenna-distributed transmitter modules. – High modularity, redundancy, and reliability. – Large ‘‘open’’ array antenna, permitting a. very good angular resolution, both in azimuth and in elevation; b. reduced emitted power with low power consumption and cooling requirements; c. good resistance to wind, adverse climatic conditions, and easy access for maintenance. – Electronic scanning in elevation. – An accurate phase/amplitude control of the antenna distribution network. – Multiple independent and contemporary beams for longer time-on-target without drawback on the track refreshing rate, allowing good azimuthal accuracy and optimum disturbance cancellation.

8.2

Phased Array Antenna

– Height measurement with excellent accuracy in all clutter environments; monopulse is often a desired feature. – Signal processing based on a multichannel configuration with automatic or manual selection of the Doppler processing for optimum performances in changing environments. – Staggered MTI filtering against stationary and/or moving clutter/chaff by using a cascade of fixed and adaptive notch MTI filters. – Constant false-alarm rate (CFAR) circuit, which prevents false-target overload due to clutter residuals. – High-quality construction and extensive built-in test equipment (BITE) for high reliability and maintainability. – Multimicroprocessor and modular software configuration, which ensure growth capability for data processing. – Secondary surveillance radar (SSR), which provides identification data within the same detection volume of the primary radar. – Primary radar plots and secondary plots are compared to provide a single associated plot. – Data processor is capable of handling a very high number of plots.

8.1.7 Selection of GBEWR Operative Frequency L-band is the preferred frequency band (instead of S-band, for example) for the operation of long-range radar [2] mainly because [22]: 1. The effect of ground, sea, and rain clutter begins to become significant as the radar frequency is increased. L-band radar takes advantage of natural reduction of clutter mean reflectivity associated with lower frequencies. At center frequency, it results in a 5-dB advantage over S-band in land clutter and a 17-dB advantage in rain clutter. 2. The atmospheric loss increases with the frequency. The L-band has typically 1 dB less attenuation than S-band at 0 of elevation and at a range of 370 km. 3. The L-band is the most suitable for defense against intercontinental BMs due, for example, to large RCS values at lower frequencies.

8.2

PHASED ARRAY ANTENNA

8.2.1 Introduction The 3-D radar is replacing the 2-D one for military applications [23–26]. The improved performances in terms of detection capability, accuracy, and flexibility with respect to the 2-D alternatives are offered at an attractive manufacturing cost, especially when compared with the operational benefits. 3-D radar permits the measurement of the complete spatial position of each target at each antenna scan. This is a distinct advantage in systems (typically, early warning, air defense, and weapon control) where the complete knowledge of target position needs to be acquired as soon as possible. Other benefits are related to large antenna aperture and spatial filtering capability. 1. The vertical antenna size is not limited by the required elevation coverage [51]. As an example, a system devoted to naval point defense purposes requires detection and

335

336

CHAPTER

8

Ground-Based Early Warning Radar (GBEWR)

measurement capability up to 80 in elevation. In 2-D architecture, this requires an antenna aperture with small vertical size, which results in an impairment of the radar detection range and height resolution. In 3-D, a narrow pencil beam can be aimed at the required elevation without reducing the antenna vertical length. 2. Larger antenna apertures permit a better control of the radiation patterns and better performances in terms of sharp cut-off at the horizon and elevation sidelobes. 3. The clutter interference is significantly reduced due to the attenuation produced by the pencil beam pattern. Rain at the upper elevation angles occurs only at short ranges: Targets not physically buried in the rain can be seen in the clear, which is not the case for 2-D. 4. The 3-D antenna pattern is affected by manmade disturbances only at small elevation angles where the interferences are impinging. Examples of these systems have been developed by Selex ES, which has been active in this field for more than 30 years, with a number of defense radars like the MRCS403 (S-band medium range), RAT31SL (S-band long range), RAT31DL, RAN 40 L (L-band long range), 743D (L-band long range), AR327 (S-band long range), AWS-9 (S-band), and their derivations (see www.selex-es.com). These radars cover both the long and medium range for air and naval defense applications. One important issue in designing 3-D radars is the selection of the beam-scanning technique (BST), which permits the acquisition of targets over the full elevation coverage in the frame time allowed by the antenna rotation speed and the azimuth beamwidth. This choice strongly determines the overall achievable performance. The L-band radars RAT31DL, RAT31DL/M, and RAN40L are the most recent members of the RAT31 family, which exploit a proprietary BST developed in SELEX Sistemi Integrati years ago and used on a number of SELEX Sistemi Integrati sensors with passive antennas (passive antennas are fed by one central high-power transmitter) [23]. Recent progress in active antennas (usually phased array antennas where, instead of a central high-power tube, every radiating element has a small power amplifier) has permitted designers to incorporate such BST into an architecture that combines the benefits of the technique with very competitive cost. In the coming years, such sensors will represent the backbone of the SELEX Sistemi Integrati NATO Class 1 product in the early warning and air defense roles, thus replacing the RAT 31SL. A number of RAT31DL radars are in service today in several countries, both within and beyond NATO borders. An evolution including ballistic missile defense (BMD) capability has been developed for the more recently acquired NATO sensors.

8.2.2 Overview of 3-D Beam-Scanning Techniques 3-D radar manufacturers mainly use three BSTs: 1. Single scanning beam: A single pencil beam (see Figure 8-3, where H stands for height and R for range; the dashed area represents the pencil beam) scans sequentially the full elevation sector to be covered both in transmission and reception. 2. Multiple stacked beams: A single transmission fan beam (see Figure 8-4 where E stands for elevation) radiates a single waveform over the whole radar coverage. N contemporary multiple pencil-stacked beams (see Figure 8-5) are used for reception over the whole elevation coverage.

8.2

Phased Array Antenna

337

3. Sequential scanning stacked beams: A single fan beam in transmission covers only a sector of the complete coverage area, and M (with M < N) contemporary multiple pencil-stacked beams cover the same sector for reception (see Figure 8-6). To explore the full coverage, both the transmission and reception beams are switched to sweep the sectors in which the full coverage has been split. SELEX Sistemi Integrati has envisaged the multiple independent scanning (MIS) beam technique is shown in Figure 8-7 [21]. On transmission, multiple pencil beams illuminate H

FIGURE 8-3 ¢ Pencil Beam.

Hmax

R

H

FIGURE 8-4 Beam.

¢

Fan

Emax Rmax

Hmax

R

H

FIGURE 8-5 ¢ Multiple Pencilstacked Beams.

R

338

CHAPTER

8

Ground-Based Early Warning Radar (GBEWR)

H

FIGURE 8-6 ¢ Sequential Scanning Stacked Beams.

4 3 2

SUBSECTOR 1 R

FIGURE 8-7 ¢ Multiple Independent Scanning Beams.

H

3

2

SUBSECTOR 1 R

M angular directions radiating in each direction a different waveform (pulse duration, code modulation, and frequency). On reception M, contemporary pencil beams acquire the signal coming from the directions where transmission occurred. The word ‘‘independent’’ is used here to indicate the following: (a) each beam can be pointed in any direction without limitation due to the pointing of other beams; and (b) the waveform to be used is not constrained to be coincident with the waveforms used in the other beams. The average power available from the transmitter is the only main issue to consider in the definition of the waveforms. In a clear environment, no meaningful difference in radar range performance is related to the adopted scanning method. Thus, the comparison has to be done: (a) in clutter conditions where the time on target (ToT) is at its premium for efficient implementation of Doppler processing; and (b) under ECM where spatial filtering and waveform diversity are determinant. 1. Single pencil beam: This configuration suffers due to time shortage. Indeed, the single beam can dwell only a very short time in each beam position. As a consequence, such radar has to: (a) transmit very long pulses to reach the required range coverage, with a consequent higher vulnerability to ARM attack and a large range

8.2

Phased Array Antenna

interval around each target where the sensitivity is reduced by the presence of sidelobes generated by the pulse compression; and (b) use large azimuthal beamwidth to gain ToT, impairing the resolution and accuracy. To recover performance, a monopulse antenna is typically used in azimuth. The wider beam width is also more susceptible to jammers. (c) Doppler processing is not performed over the whole elevation coverage. On the other hand, this approach offers: (1) high spatial filtering due to the pencil beam shape, both in transmission and reception; (2) high flexibility in choosing different waveforms for different elevations to tightly match the radar coverage envelope; and (3) simplicity—in principle, only one sum and two (in azimuth and elevation) difference channels are necessary. 2. Multiple stacked beams: This permits the longest ToT with a more efficient Doppler filtering. The use of a transmitted fan beam and a single waveform implies the following: (a) the spatial filtering capability is reduced and the same Doppler processing is applied to all beams because the clutter return is present in any beams; (b) higher detectability by ARM in the elevation coverage; (c) the lack of flexibility in choosing waveforms results in energy that is often directed where not needed; (d) the need to use many different receiving channels if a wide elevation sector has to be covered (0 –80 ); (e) lower accuracy in target elevation measurement due to the use of sum channels of adjacent beams for angle estimation (difference channels are not available); (f) reduced ECCM capability in the presence of more than one smart jammer positioned at different elevations; and (g) lack of special functionalities such as ‘‘burn through’’ or ‘‘look down’’ (valley coverage) modes.3 Many manufacturers have used this configuration in the past because, in principle, phase shifters inside the antenna are not required to point the beams. But in many cases (as on board a ship or on ground, if terrain profiling or automatic leveling corrections are required), this advantage is lost. 3. Sequential scanning stacked beams: This approach typically uses a stack of four beams switched among two or three elevation sectors, together with the corresponding transmission fan beam. The number of receiving channels is reduced; the waveform can be optimized for each sector. The fan beam covers one sector at time, so clutter in high-elevation sectors is spatially filtered with respect to low-elevation sectors. More flexibility and simplicity are achieved in the antenna, but a lower ToT is available. The drawback caused by the use of the fan beam and single waveform applies at each elevation sector. 4. Multiple independently scanned (MIS) beams: This is an extension of the single scanning pencil beam with the addition of extra independent (not stacked) beams. Due to the use of pencil beams in transmission and reception, this solution retains all the benefits of the two-way spatial filtering with the addition of a longer ToT than the single pencil beam, allowing for good Doppler filtering over the full coverage area as well as in the presence of electromagnetic anomalous propagation. The use of different waveforms and carrier frequencies with independent agility criteria per beam strongly increases the separation between the different beams, reducing the possibility of falsetarget generation and improving radar robustness against multiple smart jammers. The

3 These are operative modes required to enhance radar detection ranges in certain directions (burn through) or to point the beam at elevation angles lower than 0 (look down).

339

340

CHAPTER

8

Ground-Based Early Warning Radar (GBEWR)

transmitted waveform is shared in different directions in a sequential fashion. With MIS, each beam is tailored to the assigned task; this is derived from the ability to freely choose among the beams the pointing direction, the beamwidth, and the waveform. The most relevant benefit is achieved when the elevation scanning profile is adapted to specific operational requirements occurring in limited azimuth sectors and changing in time (e.g. TBM tracking). As an example, in the TBM tracking role, the RAT31DL is able to dedicate one beam to illuminate the TBM while the other three beams are scheduled to cover the remaining coverage area. This guarantees the maximum ToT on the missile compatible with the antenna rotation speed. This capability can be exploited in adaptive fashion, applying it in the relevant azimuthal sectors only.

8.2.3 Overview of a GBEWR Family The radar is composed of: (1) the antenna group, including a radiating array, the spine with transmit-receive modules (TRMs), an antenna cabinet with analogue receivers, and the mechanical base; (2) the equipment shelter containing the processing cabinet, the identification friend or foe (IFF), the radar environment simulator (RES), the human machine interface (HMI) and service monitors, UHF/VHF communication, modems, and time standard; and (3) the cooling unit consisting in an air cooler for the equipment shelter while the antenna group is cooled by natural air circulation.

8.2.3.1 Antenna The active antenna architecture is the key for an easy and cost-effective implementation of the MIS technique. The radiating aperture of the RAT 31DL antenna shown on the left side of Figure 8-8 [25] has 42 row planks, each one supporting the horizontal beamformer. This is a strip line power splitter, distributing the signal to the radiating dipoles with a suitable amplitude and phase to achieve a horizontal pattern with the desired beamwidth and sidelobes. Each row is connected to a TRM that combines a transmit and receive channel. Placement of power and low-noise amplifiers at the antenna aperture eliminates transmit and receive losses. A filter and a coupler are used both in transmission and reception. The filter is used both to limit the radiated spectrum and to filter out electromagnetic disturbances during reception. The receiving chain includes a duplexer, a low-noise amplifier, and a set of four independent phase and amplitude adjustment modules (PAAMs). This last component uses gallium arsenide (GaAs) monolithic microwave integrated circuit (MMIC) technology and includes radio frequency (RF) and logical interface circuits. Each PAAM is devoted to a specific beam, and all together are combined into the sum and difference beamforming networks. The amplitude control of PAAM is mainly used to recover, via calibration, the gain change of the receiving chain. The transmission contains an RF power amplifier fed by a single PAAM; the latter is sufficient because the transmitted signal consists of a cascade of FIGURE 8-8 ¢ RAT 31DL, RAT 31DL/M, and RAN 40L Antennas [Courtesy of Selex ES].

8.2

Phased Array Antenna

four different pulses that can be controlled in time sequence. A single set of PAAMs is suitable both for sum and difference signals because the pointing directions of these two beams are always made coincident. To calibrate the antenna, a test signal is injected into each TRM through a coupler, so the performance (pointing accuracy and the vertical sidelobe levels at specified values) is unaffected by large temperature variations, aging components, and component replacement. To reach low sidelobes both in transmission and reception, suitable taperings have been used, except for the elevation transmission pattern, where a two 6-dB step taper has been adopted. Figure 8-8 also shows two more Selex ES radars built with the same technology and concepts. The distributed power generation architecture has significant advantages compared to the solid-state bulk transmitter: (1) Power rotary joints are not needed. They are expensive and introduce mechanical constraints by their huge dimensions and weight. The connection between antenna group and equipment shelter is via intermediate frequency (IF) cable only because the front-end receiver is contained in the antenna cabinet, which is rotating with the antenna. This reduces cost and simplifies the cabling, permitting long connection paths. (2) The absence of insertion losses caused by power phase shifters and RF connections permits lower generated RF power. Thus, a lower number of active devices are used, a smaller prime power supply is needed, and less heat has to be removed from the array. Natural air circulation is sufficient because the dissipating surfaces on the antenna are large. (3) Growing capability: Beams can be added with a reasonable increase of modular components. Analyses have been done to compare the cost of a simple one-beam configuration with respect to an alternative four beams. There is a cost increase at the sensor level from 12 percent to 20 percent, depending on the specific application and the technology used for the receiver [26]. A relevant aspect of active array antennas is the accessibility of the components for removal and repair. This antenna is based on modular components that are easy to plug in and an elevator that is an integral part of the antenna spine, which allows for easy servicing. Furthermore, the distribution of the power source across the antenna aperture allows for graceful degradation.

8.2.3.2 TRM Technology A notional transmit-receive module consists of three main sections: transmission, reception, and auxiliary. The transmit chain consists of power amplifiers that use GaAs [27] MMIC and silicon bipolar transistors in C class and a PAAM. The receive chain consists of a P-type/intrinsic/N-type (PIN) diode limiter, a very low-noise amplifier, a 1:4 RF divider; and four PAAMs. The complete high-power section is realized with a GaAs 10-W power and 40-dB gain amplifier and suitable cascade of C-class amplifiers to provide up to 2 kW. The same modular C-class power amplifier is used as a ‘‘brick’’ in a parallel configuration to achieve all the requested power values. The brick is made of two identical transistors connected in parallel to produce more than 500 W peak with 45 percent efficiency. Typically, the PAAM (Figure 8-9) package includes hermetic multichip components with MMICs and a complete control circuit; it is plugged in a suitable number inside the TRM (typically, 4 RX plus 1 TX beam). The PAAM includes a phase shifter, an attenuator, switches, and amplifiers. The chip set, as well as the MMICs used in this system, are designed and produced using the SELEX Sistemi Integrati GaAs foundry. The metal semiconductor field effect transistor (MESFET) process was chosen based on cost and yield factors. The auxiliary section contains the power supply conditioner, a logic

341

342

CHAPTER

8

Ground-Based Early Warning Radar (GBEWR)

FIGURE 8-9 ¢ Brick and PAAM [Courtesy of SELEX Sistemi Integrati].

FIGURE 8-10 ¢ TRM [Courtesy of SELEX Sistemi Integrati].

circuit realized with field programmable gate array (FPGA) technology for self- protection, byte, and external interface. Photos of TRM are shown in Figure 8-10.

8.2.3.3 Radiating Network Technology Each radiating row plank is made of a high-power L-band, isolated divider with integrated radiating dipoles. The large antenna aperture requires the adoption of innovative technological solutions to guarantee electrical performance and mechanical robustness at a reasonable cost. Each row is subdivided into a central portion and two foldable wings. It houses 52 dipoles distributed along 11 m of length, with a weight of 25 kg and a total insertion loss > > > > 2 2 2 2 2 2 2 2 2 2 2 > > > sy ¼ sin qðsr cos j þ r sj sin jÞ þ r sq cos q cos j > > > > > s2z ¼ s2r sin2 j þ r2 s2j sin2 j > > < h i 1 2 2 2 2 2 2 2 ¼ r s Þcos j þ r s sin j s sin2q ðs xy r q j > > 2 > > > > 1 2 2 2 > > > sxz ¼ 2 cosqsin2jðsr r sj Þ > > > > > > : syz ¼ 1 sinqsin2jðs2 r2 s2 Þ r j 2

367

(8.23)

Given the state ^s k=k at the k-th time instant, with the corresponding estimation covariance matrix Pk=k , the prediction at time instant k þ 1 is ^s kþ1=k ¼ Fsk þ G½ag þ f k ðsk Þ þ gk ðsk Þ (8.24) and the covariance matrix of the predicted state is Pkþ1=k ¼ ðF þ GJ k ÞPk=k ðF þ GJ k ÞT þ Q

(8.25)

where J k (3 8 matrix) is the sum of the Jacobian of the nonlinear functions f k ðsk Þ and gk ðsk Þ calculated at the state ^s k=k estimated at the previous step. The Jacobian is J k ¼ F k þ G k ¼ ½rsk f Tk ðsk ÞT þ ½rsk gTk ðsk ÞT

(8.26)

The previously described EKF model can be used to track targets with ballistic models, even with an initially unknown ballistic coefficient, but if inserted in an IMM architecture (Figure 8-27), it provides additional advantages: 1. The radar system operates in a ‘‘dense’’ environment, i.e., in the presence of a number of different threats, like air-breathing targets (ABTs), anti-radiation missiles, BMs, and others. A different EKF filter can be designed for each type of target in the environment and all together interact via IMM. 2. BMs may in general change their dynamics as a function of the flight time; the boost is present in the first part of the trajectory and it is followed by the cruise and the reentry phases. Thus, it is required to account for the target maneuvers starting at some KF TARGET 1 w1 EKF TARGET 2 ρ, θ, f Rx

w2

SP

EKF TARGET k wk CALCULATION OF TARGET MODEL LIKELIHOOD

TARGET STATE AND TARGET MODEL

FIGURE 8-27 ¢ Typical Block Scheme of an MM Filter. Weights wk Related to the Probability of Model k given the Measurements.

368

CHAPTER

8

Ground-Based Early Warning Radar (GBEWR)

time during the estimation interval, in which case a model change occurs. The changing of the target model is controlled by the computation of the IMM mixing probabilities and the interactions among the implemented target models. 3. The BM characteristics are not generally a priori known; thus, it is required to estimate online the BM parameters to improve the track accuracies. The IMM mixes the outputs of a bank of different filters designed for different BMs, each one adapting its estimated parameters to the tracked target, thus permitting the best tracking of BMs pertaining to different ‘‘classes.’’ 4. The IMM likelihood provides for each filter in the bank a clear indication of the confidence of the tracker on the type of target under analysis; this is an intrinsic capability of non-cooperative target classification available ‘‘for free’’ by the IMM. Multiple model MM approaches (IMM, MMFM [multiple model for fixed model], and others) have an intrinsic capability of target classification. Through the MM filter development, in fact, it is possible to extrapolate classification information from the model probability of the different filters selected. The principal purpose in this case is to have some hooks into the system to deliver some preliminary ABT or BM flag and, therefore, use that early information to determine successive radar functions (i.e., imaging for refined identification). Discrimination calculations are based on the weight assigned to each filter implemented into MM. Let’s say that the filters into MM are of two different families: ●

Kalman filters (KF) to estimate the threats with kinematics typical of an ABT and more generally for highly maneuvering target, and Extended Kalman filters (EKF) designed to track threats with ballistic nonlinear kinematics (i.e., the BM in reentry phase).

8.7.1.5 Simulated Scenarios and Results Three simulated scenarios are presented here. The following radar parameters have been maintained constant for the scenarios: range accuracy ¼ 25 m, azimuth accuracy ¼ 0.15 , elevation accuracy ¼ 0.2 , and the Pd ¼ 0.9. The IMM for scenarios 1 and 3 is designed as follows: a KF with the state vector equal to ðx; x_ ; y; y_ ; z; z_ Þ matched to ABTs with constant velocity (6 degrees of freedom); a KF with the state vector equal to ðx; x_ ; €x ; y; y_ ; €y ; z; z_ ; €z Þ matched to maneuver ABTs (9 degrees of freedom); an EKF with the state vector equal to ðx; x_ ; y; y_ ; z; z_ ; bÞ matched to ‘‘multistage’’ BM (initial value of the ballistic coefficient equal to 40,000 N/m2); an EKF with the state vector equal to ðx; x_ ; y; y_ ; z; z_ ; bÞ matched to ‘‘one-stage’’ BM (initial value of the ballistic coefficient equal to 200,000 N/m2). For scenario 2, the IMM is slightly different and it will be detailed later. The BM under track is described in Section 8.8.1.3. Scenario no. 1. This simulation has the purpose of illustrating the working principle of the IMM showing the update probability model of a tracking architecture constituted by two KFs: The first KF has been designed for ABT with a constant velocity, while the second KF is pertinent to a maneuvering ABT. The simulated ABT trajectory is shown in Figure 8-28 in the xy plane, while the IMM probabilities are reported in Figure 8-29. In Figure 8-29 the transitions from one filter to the other are shown: The IMM selects the filter matched to the ABT kinematics, i.e., the update probability of the KF matched to

Special Functions

8.7 100

369 FIGURE 8-28 ¢ ABT Reference Trajectory.

90 80

y (km)

70 60 50 40 30 20 Radar in the Axis Origin 10

40

50

60

70 x (km)

80

90

100

Mixing Probability 1

KF (6 dim)

0.9

KF (9 dim)

0.8

KF (9 dim)

KF (6 dim)

0.7 Update Probability

KF (9 dim)

KF (6 dim) a = 10 sa = 0.5 KF (9 dim) a = 20 sa = 0.5 EKF a = 10 sa = 0.5b = 40,000 EKF a = 10 sa = 0.5 b = 200,000

0.5 0.4

KF (6 dim)

KF (6 dim)

0.6

KF (6 dim)

KF (9 dim)

EKF (TBM)

0.3 0.2 0.1 0

50

100

150

200

250

300

350

400

450

t (s)

the ABT model is practically equal to 1, while the analogous probability of the filters mismatched to the ABT kinematics is practically 0. Scenario no. 2. This scenario has been considered to demonstrate the benefit of including a suitable boost acceleration filter in the IMM. The BM described in Section 8.8.1.3 (see also Figure 8-25) is adopted in the simulation. Figure 8-30 presents the

FIGURE 8-29 ¢ Update Probability of the IMM Filters. (KF: Kalman filter, EKF: Extended Kalman Filter, b ¼ b ¼ Ballistic Coefficient Starting Value, a ¼ Reciprocal of the Expected Target Maneuver Time Duration, sa ¼ Standard Deviation of Acceleration).

FIGURE 8-30 ¢ Mean Value and Standard Deviation Along z-axis; No boost filter used.

CHAPTER

8

Ground-Based Early Warning Radar (GBEWR)

500 0 Radar Measurement

–500 Ez (m)

370

–1,000 –1,500 EKF Estimation

–2,000 –2,500 –3,000

50

100

150

200

250

300

250

300

t (s)

End of Boost 600 500

Radar Measurement

sz (m)

400 300 200 100 0

EKF Estimation 0

50

100

150

200

t (s)

500

FIGURE 8-31 ¢ Boost Filter Included in the IMM.

0 Ez (m)

–500 –1000 –1500 IMM Estimation

–2000 –2500 –3000

50

Radar Measurement

100

150 t (s)

200

250

300

250

300

600 Radar Measurement

500 sz (m)

400 300 200 100 0

IMM Estimation 50

100

150 t (s)

200

accuracy results along the z-axis achieved with an EKF designed for a cruise and re-entry phases of the BM. The upper curves are related to the mean value of the position error; lower curves are related to the error standard deviation. The EKF is not matched to the boost phase of BM; thus, the error in the mean value is evident. Figure 8-31 refers

Special Functions

8.7

371

to the same simulation, but the tracker is an IMM with two EKFs; the first is matched to the BM cruise and re-entry phases, while the second is designed for the boost phase; the benefit of the boost phase filter appears in the reduced mean value of the tracker error. For this scenario, it is assumed an a priori knowledge of the BM characteristics.

500 0 –500 –1,000 –1,500 –2,000 –2,500 –3,000 65

sz (s)

Ez (m)

Scenario no. 3. This simulation shows the capability of the IMM architecture to be adaptive to the unknown BM characteristics. Consider the case of absence of knowledge of b, Isp, and the other BM parameters of the BM under tracking. The IMM architecture is constituted by a number of EKFs matched to the BM dynamics (see Section 8.8.1.4), each filter having the capability of estimating online the BM characteristics (for instance, the b as a value). Figure 8-32 reports the mean value and the standard deviation of the estimated error along the z-axis before and after the tracker filtering for the BM presented in Figure 8-25. For the sake of simplicity, only the cruise and re-entry phases of the BM have been filtered. Note that the IMM is capable of tracking the BM by adaptively adjusting the EKF parameters during the BM flight time. The accuracy reported along the z-axis of Figure 8-32 is similar to the same quantity presented in Figure 8-31 that was achieved, assuming a perfect knowledge of the ballistic target characteristics. The lesson learned is to design EKFs filter matched to ‘‘general’’ BM motion equations and develop a tracking architecture based on the IMM that allows the user to maintain under track BMs with unknown ‘‘kinematic parameters.’’ It is also important to note that the conceived tracking architecture has an intrinsic classification capability. Figure 8-33 reports the internal behavior of the four filters previously described and includes one KF with 6 degrees of freedom, one KF with 9 degrees of freedom, and two EKFs with initial b value equal to b in the picture (while a is related to the injected plant noise). The internal behavior is fully described by the mixing probability: In this case, the EKFs have been selected after few seconds from the track initialization, and the confidence of the IMM is practically equal to 1. This constitutes reliable information about the type (either BM or ABT) of target under tracking.

IMM Estimation Radar Measurement

100

150

t (s)

600 500 400 300 200 100 0 65

FIGURE 8-32 ¢ Accuracy of the IMM in the Presence of an Unknown BM.

200

250

300

250

300

Radar Measurement

IMM Estimation 100

150

t (s)

200

FIGURE 8-33 ¢ Update Probability of the IMM (KF: Kalman Filter, EKF: Extended Kalman Filter, b ¼ b ¼ Ballistic Coefficient Starting Value, a ¼ Reciprocal of the Expected Target Maneuver Time Duration, sa ¼ Standard Deviation of Acceleration).

CHAPTER

8

Ground-Based Early Warning Radar (GBEWR)

1 0.9 0.8

KF (6 dim) a = 0.5sa = 10 KF (9 dim) a = 0.5sa = 20 EKF a = 0.5sa = 10 b = 40,000 EKF a = 0.5sa = 10 b = 20,000

0.7 Update Probability

372

0.6 0.5 0.4 Start the drag 0.3 0.2 0.1 0 65

100

150

200

250

300

t (s)

8.7.1.6 Anti-Radar Missile (ARM) Detection and Alert GBEWR exploits several techniques to prevent detection by an intercept receiver, thus making it difficult for an ARM to lock on the radar emission. 1. The used transmitted waveform spreads the energy spectrum of the radar over the widest possible band. The use of a pseudo-random code prevents the intercept receiver from performing a matched filter to the radiated waveform. In fact, the transmitted frequency of the GBEWR can be selected either in a random fashion or under the control of the AFS device, among, say, hundreds frequency values within hundreds of MHz in the radar band. The GBEWR also adopts irregular modulation types of the transmitted waveform, such as pulse trains with nonuniform and staggered PRFs; it also uses variable dwell time and it changes the waveforms during the operation. 2. Most intercept receivers are designed to detect short radar pulses; coherence between radar pulses in a pulse train is generally ignored. Also, there is a little capability against pulse compression or noise-like radar waveforms. The time-bandwidth product of the radar code within the transmitted pulse should be as large as possible. In fact, this product expresses the relative mismatch between the transmitted radar waveform and the intercept receiver. 3. An ultra-low sidelobe antenna reduces the radiation level through the radar’s sidelobes. This follows, since most ARMs are designed to attack through the radar’s sidelobes where the signal is continuously available. The GBEWR looks for sidelobe level better than 45 dB down to the peak of the antenna main beam, thus making it difficult for an intercept receiver to detect and locate the EM radiation source.

8.7

Special Functions

4. A narrow beam in azimuth and in elevation reduces the volume of the space where the radar energy is radiated. This is important against that ARM that is sensitive to the flashes of energy radiated by the radar mainbeam. The lower beam in elevation of the GBEWR, which is the one that may be intercepted by a far ESM (electronic support measures) platform from which the ARM could be launched, has typical width values around 2 for the azimuth and elevation, respectively. 5. The GBEWR has a low-peak power combined with a multiple beam in elevation. This is advantageous with respect to another system solution built around a single beam with a high-peak power transmitter. 6. A typical GBEWR may have an elevation coverage up to, say, 30 . This means that an ARM that is approaching the radar flying, say, at 20,000 m altitude, is within the radar coverage up to approximately 60 km from the radar site. This means that the GBEWR has many opportunities to detect the ARM before the threat penetrates within the zenithal hole region above the radar where its detection capability is minimal. 7. When a track is radial toward the radar and a suspect rises that the ARM is locking on to the radar, then the natural defense of GBEWR is simply either to turn off the transmission or to activate suitably located transmitting decoys. In fact, any interruption of the radar’s radiations leads to a complete loss of guidance information to the ARM.

8.7.2 Low Probability of Intercept (LPI) It is generally accepted [81] that a quantitative measure of the radar efficiency in the presence of an ESM receiver is computed via the following equation: 2 2 2 2 1 ET TR 1 LR l SR ðnÞ GTI GI a4 ¼ (8.27) 4p hI tBI TI s GT GR SI2 ð1Þ L2I where a ¼ RI/Rm, RI ¼ ESM detection range, Rm ¼ radar detection range, ET ¼ energy transmitted by the radar, hI ¼ ESM system noise spectral density, i.e., KTI, K ¼ Boltzman constant, TI ¼ ESM system noise temperature, TR ¼ radar system noise temperature, t ¼ radar pulse time duration, BI ¼ ESM bandwidth, LR ¼ radar losses, LI ¼ ESM losses, l ¼ radar wavelength, s ¼ target radar cross section, SR(n) ¼ single-pulse useful signal to system noise power ratio (SNR) for the radar having fixed Pd (detection probability), Pfa (false alarm probability), and n pulses to be integrated,

373

374

CHAPTER

8

Ground-Based Early Warning Radar (GBEWR)

SI (1) ¼ single-pulse SNR for the interceptor having fixed Pd and Pfa, GTI ¼ radar antenna gain along the interceptor direction of arrival, GI ¼ interceptor antenna peak gain, GT ¼ radar antenna transmission peak gain, and GR ¼ radar antenna receiving peak gain. A value for a lower than 1 indicates an advantage for the radar being able to detect the ESM at a higher distance with respect to the ESM detection range being the radar the target to be revealed. If a is higher than 1, then the ESM has an advantage with respect to the radar. The distance corresponding to a ¼ 1 is called the quiet range. For ranges higher than the quiet range (a > 1), the ESM detects the radar before the radar detects the ESM; on the contrary, before the quiet range (a < 1), the radar detects the ESM before being detected by the ESM.

8.7.2.1 Discussion on the Evaluation of the Radar LPI Capability Equation (8.27) has been obtained by combining the two equations that give the SNR value in correspondence of the radar antenna (produced by the echo reflected by the ESM) and the SNR in correspondence of the ESM antenna (produced by radar emission). The two equations are SR ðnÞ ¼

PT GT s GR l2 1 t 4pR2m 4pR2m 4p LR KTR

PT GI l2 1 1 SI ð1Þ ¼ G TI 4p LI KTI BI 4pR2I

(8.28)

The following considerations are in order: 1. Radars transmitting very high peak power are most easily detectable by the ESMs because ET increases a. 2. The lower the ESM RCS, the higher the radar difficulty to detect it; in fact, s increases a. RCS equal to 1 m2 is generally selected for the evaluation. 3. GTI represents the radar antenna gain along the ESM direction of arrival. If the ESM is in the antenna mainbeam, then GTI is the radar antenna peak gain usually determined by other factors than the LPI capability. If the interceptor is in the antenna sidelobes, then GTI is the sidelobes’ value itself. Low-antenna sidelobes are clearly advantageous. 4. The (t BI) product is a measure of the matching between the radar waveform and the receiver bandwidth. The radar objective is to increase the product as much as possible to minimize a; the ESM objective is t BI ¼ 1. For the purpose of this section, BI is approximated following the procedure (assuming a square law detection) suggested in [81]: pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi BI ¼ 2BRF BV (8.29) where BRF is called ‘‘pre-detection bandwidth’’ and BV ‘‘post-detection bandwidth.’’ If we hypothesize that the ESM does not exactly know the radar carrier frequency, then BRF is considerably larger than the minimum required. BV is, in general, selected such that very short pulses can be transmitted (tmin); a common

8.7

Special Functions

375

choice is BV ¼ 0.5/tmin. A mismatching factor is defined as M ¼ t/tmin, and the ESM bandwidth becomes pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi BI ¼ MBRF =t (8.30) After all, (t BI) can be rewritten as ðtBI Þ2 ¼ MBRF t ¼ M 2 BRF tmin

(8.31)

Let us consider the case of a coded pulse. The ESM computes the envelope of the radar signal, regardless of the coded modulation; this means that the coded pulse appears to the ESM as a longer uncoded pulse. Let us denote by b the compression ratio of the coded pulse fired by the radar and by MP ¼ tEFF/tmin the ratio between the time duration of the compressed pulse (tEFF ) and tmin defined earlier; then the product (t BI)2 is equal to ðtBI Þ2 ¼ b2 Mp2 BRF tmin From this new expression of the t BI product, it stems that a is inversely proportional to b; the radar should have very large b to be effective from the LPI perspective. In conclusion, we can claim that a radar firing a long coded pulse with low peak power is advantageous for the ‘‘radar quietness’’; this generally occurs in the radar of the class described in this chapter.

8.7.3 Denial of Bistatic Hosting by Waveform Design Many countries have heavily invested in the development of advanced surveillance systems and technologies. An increasing concern is that potential adversaries may use bistatic technologies to gain capability versus significant investments in advanced sensors [82]. Indeed, with relatively inexpensive receiver systems, an adversary can use other signals as bistatic ‘‘illuminators of opportunity.’’ The bistatic radar needs to acquire the illuminator signal and to correlate it as coherent reference signal versus the received signals acquired on the targets directions of arrival (TDoA). As illustrated in Figure 8-34, a coherent reference is typically obtained by measuring a direct path signal via the sidelobes of the illuminator [83]. Target Transmit Mainbeam Monostatic Radar Bistatic Receiver Mainbeam Spatially Denied Coherent Reference

vvvvvv

Non cooperative Bistatic Sensor

FIGURE 8-34 ¢ Noncooperative Bistatic Receivers Require Coherent Reference from the Host Illuminator.

376

CHAPTER

8

Ground-Based Early Warning Radar (GBEWR)

FIGURE 8-35 ¢ Radar and Masking Signal Radiation.

Radar Signal Pattern Fr(θ) Masking Signal Pattern Fm(θ)

–90°

+90°

Azimuth Angle θ

If conventional methods to prevent the interception of the direct path signal include low sidelobe antennas, physical isolation, and the use of spread spectrum waveforms, in [82], a number of theoretical techniques have been introduced and evaluated to prevent a radar being used by an adversary as a bistatic illuminator of opportunity. All of these techniques are based on the idea of radiating a so-called ‘‘masking signal’’ (Figure 8-35), which is arranged to be approximately orthogonal, both in a spatial sense and in a coding sense, to the radar signal, and on the idea of a sufficient level to mask the radar signal to an adversary, and hence, to deny a reference for bistatic operation. Therefore, the solution needs to find a radar waveform ur(t) with suitable ambiguity function and a masking waveform um(t) that is approximately orthogonal to the radar waveform over the full range and Doppler domain. So if the radar waveform is radiated via a pattern Fr(q) and the masking waveform um(t) is radiated via a pattern Fm(q), then Fr(q) and Fm(q) need to be approximately spatially orthogonal over the full bandwidth of the radar. A number of waveform coding techniques have been analyzed in [82], and among those considered, Costas codes appear to offer the best performance and flexibility. Two spatial coding techniques have been devised and analyzed—one based on an interferometer, and one based on a Butler matrix. Expressions as a function of the system parameters have been derived for the degree of hiding of the radar signal by the masking signal, and for the suppression of the masking signal in the host radar echo. Evaluation and plotting of these expressions have demonstrated that it is possible to obtain adequate masking of the radar signal, while at the same time achieving suppression of echoes from the masking signal of the order of 30 or 40 dB: with respect to this capability, the performance given by the interferometer and by Butler matrix schemes are comparable.

8.8

CONCLUSIONS AND FURTHER READING

In this chapter, we have described GBEWR, providing some historical, theoretical, and technological details. Specifically, we first introduced the concept of GBEWR and emphasized its importance. Then, we traced back the history of EWR up to the current days, also providing an overview of some commercial products. Then, we focused on transmitter-receiver as well as antenna technology, with an emphasis on the SELEX Sistemi Integrati state of the art.

8.9

References

We also discussed advanced signal and data processing techniques exploited in GBEWR—for instance, the important function of BM tracking. For further reading, the authors recommend some classic textbooks on the story of EWR and in general radar systems such as [84–87]. The chapter is focused on GBEWR, and in order to complete an understanding of EWR, a text on airborne EWR is necessary. In this context, we recommend [88]. In addition, an overview of OTHR is fundamental and, to this end, we suggest [9]. Finally, the seminal paper [89] represents an interesting contribution to the topic of upgrading obsolete GBEWRs for homeland security protection.

8.9

REFERENCES

[1] Wikipedia, http://en.wikipedia.org/wiki/Early_warning_radar. [2] M. Skolnik, Radar Handbook, McGraw-Hill, 3rd Ed., 2008. [3] R. W. Sturdevant, ‘‘Radar, Long-Range Early Warning Systems,’’ Encyclopedia of 20thCentury Technology, http://www.bookrags.com/tandf/radar-long-range-early-warning-tf/. [4] H. Griffiths and N. Willis, ‘‘Klein Heidelberg – The First Modern Bistatic Radar System,’’ IEEE Trans. on Aerospace and Electronic Systems, Vol. 46, No. 4, pp. 1571–1588, October 2010. [5] S. N. Stitzer, ‘‘Reverberations – The SCR-270 Radar,’’ IEEE Microwave Magazine, Vol. 8, pp. 88–98, June 2007. [6] E. Brookner, Practical Phased Array Antenna Systems, Artech House, 1991. [7] D.H. Sinnott, ‘‘The Development of Over-the-Horizon Radar in Australia,’’ 1988, http:// www.dsto.defense.gov.au/attachments/The_development_of_over-the-horizon_radar.pdf. [8] D. H. Sinnott, ‘‘Defense Radar Development in Australia: 1939 to the Present,’’ IEEE Intern. Radar Conference, pp. 5–9, 2005. [9] G. A. Fabrizio, High Frequency Over-the-Horizon Radar – Fundamental Principles, Signal Processing, Practical Applications, McGraw-Hill Professional, San Francisco, CA, 2013. [10] ‘‘NADGE’’, http://www.namsa.nato.int/gallery/ws_nadge_e.htm. [11] C. P. Satterthwaite, ‘‘Space Surveillance and Early Warning Radars: Buried Treasure for the Information Grid,’’ http://www.dtic.mil/cgi-bin/GetTRDoc?Location=U2&doc= GetTRDoc.pdf&AD=ADA468199. [12] A. Odile, ‘‘Modular Design Principles for Missile Defense Early Warning Radar,’’ International Radar Conference - Surveillance for a Safer World, October 2009. [13] http://www.selex-es.com/IT/Common/files/SelexSI/brochure_datasheet/2008/Data_Sheet/ RAT_31DL.pdf. [14] http://www.thalesraytheon.com/fileadmin/tmpl/Products/pdf/111058_GM400_ Paris_Update_v3_LR.pdf. [15] http://www.lockheedmartin.com/data/assets/7152.pdf. [16] JANE’S: www.janes.com (Defense Equipment and Technology Section). [17] http://www.maquina-de-combate.com/articulo/3d-lanza-family-radars.pdf. [18] http://www.baesystems.com/BAEProd/groups/public/documents/bae_publication/038035.pdf.

377

378

CHAPTER

8

Ground-Based Early Warning Radar (GBEWR)

[19] http://www.selex-es.com/IT/Common/files/SelexSI/brochure_datasheet/datasheet_terrestri/ LR/RAT31DLM.pdf. [20] http://www.raytheon.com/capabilities/rtnwcm/groups/ncs/documents/content/ rtn_ncs_products_masterm_pdf.pdf. [21] http://www.lockheedmartin.com/data/assets/8408.pdf. [22] R. J. Galejs, ‘‘Volume Surveillance Radar Frequency Selection,’’ Proc. of IEEE Intern. Radar Conference, Alexandria (Va), pp. 187–192, 7–12 May 2000. [23] B. Palumbo, ‘‘Some Example of Systems Developments in Italy Based on Phased-Array Technology,’’ IEEE Intl. Symposium on Phased Array Systems and Technology, Boston (Ma), pp. 444–449, 15–18 October 1996. [24] M. Cicolani, A. Farina, E. Giaccari, F. Madia, R. Ronconi, and S. Sabatini ‘‘Some Phased Array Systems and Technologies in AMS,’’ IEEE Intl. Symposium on Phased Array Systems and Technology, Boston (Ma), 14–17 October 2003. [25] A. Fiorello, M. Grossi, and S. Pagliai, ‘‘Advances in Radar Systems by SELEX Sistemi Integrati: Today and Towards the Future,’’ 6th European Radar Conference (EuRAD 2009), Rome, Italy, 30 September–2 October 2009. [26] A. Cetronio, M. D’Urso, A. Farina, A. Fiorello, L.Timmoneri, and M. Teglia, ‘‘Phased Array Systems and Technologies in SELEX-Sistemi Integrati: State Of Art and New Challenges,’’ Invited Aasplenary talk, Phased-Array 2010 Symposium, Boston (USA). October 2010. [27] ‘‘ITRS International Technology Roadmap for Semiconductors – System Drivers,’’ 2007, http://www.itrs.net/links/2007itrs/2007_chapters/2007_SystemDrivers.pdf. [28] M. A. Richards, J. A. Scheer, and W. A. Holm, ‘‘Principles of Modern Radar: Basic Principles,’’ Vol. 1, SciTech, May 2010. [29] C. E. Cook and M. Bernfield, Radar Signals: An Introduction to Theory and Application, New York, Academic Press, 1967. [30] D. F. DeLong, and E. M. Hofstetter, ‘‘On the Design of Optimum Radar Waveforms for Clutter Rejection,’’ IEEE Trans. on Inform. Theory, Vol. 13, No. 3, pp. 454–463, July 1967. [31] C. Stutt, and L. Spafford, ‘‘A Best Mismatched Filter Response for Radar Clutter Discrimination,’’ IEEE Trans. on Inform. Theory, Vol. 14, No. 2, pp. 280–287, March 1968. [32] Y. I. Abramovich, and M. B. Sverdlik, ‘‘Synthesis of a Filter Which Maximizes the Signalto-Noise Ratio Under Additional Quadratic Constraints,’’ Radio Eng. Electron. Phys., Vol. 15, No. 11, pp. 1977–1984, November 1970. [33] V. T. Dolgochub, and M. B. Sverdlik, ‘‘Generalized-Filters,’’ Radio Eng. Electron. Phys., Vol. 15, pp. 147–150, January 1970. [34] P. Stoica, J. Li, and M. Xue, ‘‘Transmit Codes and Receive Filters for Radar,’’ IEEE Signal Processing Magazine, Vol. 25, No. 6, pp. 94–109, November 2008. [35] S. D. Blunt, and K. Gerlach, ‘‘Adaptive Pulse Compression via MMSE Estimation,’’ IEEE Trans. on Aerospace and Electronic Systems, Vol. 42, No. 2, pp. 572–584, April 2006. [36] Y. I. Abramovich, and M. B. Sverdlik, ‘‘Synthesis of Filters Maximizing the Signal-toNoise Ratio in the Case of a Minimax Constraint on the Sidelobes of the Crossambiguity Function,’’ Radio Eng. Electron. Phys., Vol. 16, pp. 253–258, February 1971. [37] S. Zoraster, ‘‘Minimum Peak Range Sidelobe Filters for Binary Phase-Coded Waveforms,’’ IEEE Trans. on Aerospace and Electronic Systems, Vol. 16, No. 1, pp. 112–115, January 1980.

8.9

References

[38] H. M. Finn, ‘‘Adaptive Detection in Clutter,’’ Proc. of the National Electronics Conference, Vol. 22, pp. 562–567, 1966. [39] B. O. Steenson, ‘‘Detection Performance of a Mean Level Threshold,’’ IEEE Trans. on Aerospace and Electronic Systems, Vol. 4, No. 4, pp. 529–534, July 1968. [40] H. M. Finn, and R. S. Johnson, ‘‘Adaptive Detection Mode with Threshold Control as a Function of Spatially Sampled Clutter-Level Estimates,’’ RCA Review, Vol. 29, pp. 414–464, September 1968. [41] R. Nitzberg, ‘‘Constant-False-Alarm-Rate Signal Processors for Several Types of Interference,’’ IEEE Trans. on Aerospace and Electronic Systems, Vol. 8, No. 1, pp. 27–44, January 1972. [42] V. G. Hansen, and J. H. Sawyers, ‘‘Detectability Loss Due to Greatest-Of-Selection in a Cell Averaging CFAR,’’ IEEE Trans. on Aerospace and Electronic Systems, Vol. 16, No. 1, pp. 115–118, January 1980. [43] G. V. Trunk, ‘‘Range Resolution of Targets Using Automatic Detectors,’’ IEEE Trans. on Aerospace and Electronic Systems, Vol. 14, No. 5, pp. 750–755, September 1978. [44] H. Rohling, ‘‘Radar CFAR Thresholding in Clutter and Multiple Target Situations,’’ IEEE Trans. on Aerospace and Electronic Systems, Vol. 19, No. 4, pp. 608–621, July 1983. [45] P. P. Gandhi, and S. A. Kassam, ‘‘Analysis of CFAR Processors in Nonhomogeneous Background,’’ IEEE Trans. on Aerospace and Electronic Systems, Vol. 24, No. 4, pp. 427– 445, July 1988. [46] J. T. Rickard, and G. M. Dillard, ‘‘Adaptive Detection Algorithms for Multiple Target Situations,’’ IEEE Trans. on Aerospace and Electronic Systems, Vol. 13, No. 4, pp. 338– 343, July 1977. [47] H. Goldman, and I. Bar-David, ‘‘Analysis and Application of the Excision CFAR Detector,’’ IEE Proc. on Radar and Signal Processing, Vol. 135, No. 6, pp. 563–575, December 1988. [48] E. Conte, M. Longo, and M. Lops, ‘‘Two-Sided Censored Mean-Level Detector for CFAR in Multiple Target Situations and Clutter Edges,’’ Alta Frequenza, Vol. LVIII, No. 2, pp. 165–173, March–April 1989. [49] S. D. Himonas, and M. Barkat, ‘‘Automatic Censored CFAR Detection for Nonhomogeneous Environments,’’ IEEE Trans. on Aerospace and Electronic Systems, Vol. 28, No. 1, pp. 286–304, January 1992. [50] A. De Maio, A. Farina, and G. Foglia, ‘‘Design and Experimental Validation of Knowledge-Based Constant False Alarm Rate Detectors,’’ Proc. IET Radar, Sonar and Navigation, Vol. 1, No. 4, pp. 308–316, August 2007. [51] D. K. Barton, Modern Radar System Analysis, Artech House, 1988. [52] M. Skolnik, Radar Handbook, McGraw-Hill, 2nd Ed., 1990, page 20.14, eqn 20.2. [53] A. Farina, and F.A. Studer, Radar Data Processing. Introduction and Tracking, Vol. 1, Researches Studies Press., England (Editor P. Bowron), John Wiley & Sons (USA), May 1985. Translated into Russian (Radio I Sviaz Moscow in 1993) and into Chinese (China Defense Publishing House in 1988). [54] H. Blom and Y. Bar-Shalom, ‘‘The Interacting Multiple Model Algorithm for Systems with Markovian Switching Coefficients,’’ IEEE Trans. on Automatic Control, Vol. AC-33, No. 8, pp. 780–783, August 1988.

379

380

CHAPTER

8

Ground-Based Early Warning Radar (GBEWR)

[55] X. R. Li, ‘‘Engineer’s Guide to Variable-Structure Multiple-Model Estimation for Tracking. Ch. 10,’’ http://ece.engr.uno.edu/isl/PublicationsBySubjects.htm. [56] M. De Feo, A. Graziano, R. Miglioli, and A. Farina, ‘‘IMMJPDA Versus MHT and Kalman Filter with NN Correlation: Performance Comparison,’’ IEE Proc. on Radar, Sonar and Navigation (Pt. F), Vol. 144, No. 2, pp. 49–56, April 1997. [57] R. Graziano, R. Miglioli, and A. Farina, ‘‘Multiple Hypothesis Tracking vs. Kalman Filter with Nearest Neighbor Correlation. Performance Comparison,’’ AGARD MSP 3rd Symposium on ‘‘Tactical Aerospace C3I in Coming Years,’’ Lisbon, Portugal, 15–18 May 1995, published in CP-557, pp. 25–1, 25–11 [58] R. Torelli, A. Graziano, and A. Farina, ‘‘IM3HT Algorithm: A Joint Formulation of IMM and MHT for Multitarget Tracking,’’ Invited Paper, Proc. of European Control Conference, ECC97, Vol. 3 ‘‘Estimation,’’ Pt.1 WE-A F1, p. 750, Bruxelles, 1–4 July 1997. [59] R. Torelli, A. Graziano, and A. Farina, ‘‘IM3HT Algorithm: A Joint Formulation of IMM and MHT for Multi-Target Tracking,’’ European Journal of Control, Vol. 5, pp. 46–53, 1999. [60] A. Farina, P. Lombardo, and M. Marsella, ‘‘Joint Tracking and Identification Algorithms for Multisensor Data,’’ IEE Proc. on Radar, Sonar and Navigation, Vol. 149, No. 6, pp. 271–280, December 2002. [61] L. Maisel, ‘‘Performance of Sidelobe Blanking System,’’ IEEE Trans. on Aerospace and Electronic Systems, Vol. 4, No. 2, pp. 174–180, March 1968. [62] A. Farina, Antenna-Based Signal Processing Techniques for Radar Systems, Artech House, Boston, 1992. [63] D. H. Harvey, and T. L. Wood, ‘‘Designs for Sidelobe Blanking Systems,’’ Proc. of the IEEE Int. Radar Conference, Washington D.C., USA, pp. 410–416, April 1980. [64] M. R. O’Sullivan, ‘‘A Comparison of Sidelobe Blanking Systems,’’ Proc. of the IEE Int. Radar Conference, London, UK, pp. 345–349, April 1987. [65] A. Farina, and F. Gini, ‘‘Calculation of Blanking Probability for the Sidelobe Blanking for Two Interference Statistical Models,’’ IEEE Signal Processing Lett., Vol. 5, No. 4, pp. 98– 100, April 1998. [66] D. A. Shnidman, and S. S. Toumodge, ‘‘Sidelobe Blanking with Integration and Target Fluctuation,’’ IEEE Trans. on Aerospace and Electronic Systems, Vol. 38, No. 3, pp. 1023– 1037, July 2002. [67] A. Farina and F. Gini, ‘‘Interference Blanking Probabilities for SLB in Correlated Gaussian Clutter Plus Noise,’’ IEEE Trans. on Signal Processing, Vol. 48, No. 5, pp. 1481–1485, May 2000. [68] A. De Maio, A. Farina, and A. Gini, ‘‘Performance Analysis of the Sidelobe Blanking System for Two Fluctuating Jammer Models,’’ IEEE Trans. on Aerospace and Electronic Systems, Vol. 41, No. 3, pp. 1082–1091, July 2005. [69] P. Swerling, ‘‘Radar Probability of Detection for Some Additional Fluctuating Target Cases,’’ IEEE Trans. on Aerospace and Electronic Systems, Vol. 33, No. 2, pp. 698–709, April 1997. [70] A. De Maio, A. Farina, and G. Foglia, ‘‘Target Fluctuation Models and Their Applications to Radar Performance Prediction,’’ IEE Proc. on Radar, Sonar and Navigation, Vol. 151, No. 5, pp. 261–269, October 2004. [71] A. Farina, L. Timmoneri, and R. Tosini, ‘‘Cascading SLB and SLC Devices,’’ Journal Signal Processing, Vol. 45, No. 2, August 1995.

8.9

References

[72] Y. Bar-Shalom, X. R. Li, and T. Kirubarajan, Estimation with Applications to Tracking and Navigation, John Wiley & Sons, 2001. [73] R. D. Graglia, P. L. E. Uslenghi, R. Vitiello, and U. F. D’Elia, ‘‘Electromagnetic Scattering for Oblique Incidence on Impendance Bodies of Revolution,’’ IEEE Trans. on Antennas and Propagation, Vol. 43, No. 1, pp. 1126, January 1995. [74] U. F. D’Elia, M. G. Del Gaudio, and D. Pistoia, ‘‘Electromagnetic Modelling of Complex Targets in the Near Zone and Comparison with Experimental Results,’’ IEE Int. Radar Conference, Edinburgh, 14–16 October 1997. [75] U. F. D’Elia, ‘‘Target Modelling for High Resolution Radar in the New Wide Band Seeker Applications,’’ Multinational Ballistic Missile Defense (BMD) Conference, Dallas, Texas USA, 3–6 June 2002. [76] P. Ya. Ufimtsev, ‘‘Fundamentals of the Physical Theory of Diffraction,’’ 2007, Reviews and Abstracts Antennas and Propagation Magazine, IEEE, Vol. 50, No. 1, pp.159–161, February 2008. [77] P. Zarchan, Tactical and Strategic Missile Guidance, 3rd ed. AIAA Inc., 1997. [78] A. Farina, B. Ristic, and D. Benvenuti, ‘‘Tracking a Ballistic Target: Comparison of Several Non Linear Filters,’’ IEEE Trans. on Aerospace and Electronic Systems, Vol. 38, No. 3, pp. 854–867, July 2002. [79] B. Ristic, A. Farina, D. Benvenuti, and S. Arulampalam, ‘‘Performance Bounds and Comparison of Non-Linear Filters for Tracking a Ballistic Object on Re-Entry,’’ IEE Proc. on Radar, Sonar and Navigation, Vol. 150, No. 2, pp. 65–70, 2003. [80] A. Farina, M. G. Del Gaudio, U. D’Elia, S. Immediata, L. Ortenzi, L. Timmoneri, and M. R. Toma, ‘‘Detection and Tracking of Ballistic Target,’’ IEEE Radar Conference, Philadelphia (USA), pp. 26–29, April 2004. [81] D. C. Schleher, ‘‘Low Probability of Intercept Radar,’’ IEEE Int. Radar Conference, Arlington (VA), pp. 346–349, May 6–9, 1985. [82] D. C. Schleher, Introduction to Electronic Warfare, Artech House, p. 59, 1986. [83] S. Ertan, H. D. Griffiths, M. C. Wicks, P. Antonik, D. Weiner, R. Adve, and I. Fotinopoulos, ‘‘Bistatic Radar Denial by Spatial Waveform Diversity,’’ IEE Int. Radar Conference, Edinburgh UK, pp. 17–21, October 15–19, 2002. [84] H. D. Griffiths, and S. M. Carter, ‘‘Provision of Moving Target Indication in an Independent Bistatic Radar Receiver,’’ The Radio and Electronic Engineer, Vo1. 54, No. 7/8, pp. 336–342, July/August 1984. [85] C. Dobinson, Building Radar: Forging Britain’s Early-Warning Chain, 1935-1945, Methuen Pub Ltd, January 2008. [86] R. C. Watson, Jr., Radar Origins Worldwide: History of Its Evolution in 13 Nations Through World War II, Trafford Publishing, November 2009. [87] R. Morenus, Dew Line: Distant Early Warning, The Miracle of America’s First Line of Defense, Rand McNally, New York, 1957. [88] R. Wells, Early Warning: Electronic Guardians of Our Country, Prentice-Hall, 1962. [89] G. W. Stimson, Introduction to Airborne Radar, Second Edition, Scitech Publishing, USA, 1998.

381

CHAPTER

Surface Moving Target Indication

9

William L. Melvin, Ph.D., Georgia Institute of Technology, Georgia Tech Research Institute, Atlanta, GA, USA

Chapter Outline 9.1 9.2 9.3 9.4 9.5 9.6 9.7 9.8 9.9 9.10 9.11 9.12

Introduction . . . . . . . . . . . . . . . . . . . . . . . . SMTI Radar Operation . . . . . . . . . . . . . . . . Signal Models. . . . . . . . . . . . . . . . . . . . . . . SMTI Metrics . . . . . . . . . . . . . . . . . . . . . . . Antenna and Waveform Considerations . . . Clutter-Mitigation Approaches . . . . . . . . . . Detection Processing . . . . . . . . . . . . . . . . . Angle and Doppler Estimation . . . . . . . . . . Other Considerations . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . Further Reading . . . . . . . . . . . . . . . . . . . . . References. . . . . . . . . . . . . . . . . . . . . . . . .

9.1

........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........

....... ....... ....... ....... ....... ....... ....... ....... ....... ....... ....... .......

........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........

.... .... .... .... .... .... .... .... .... .... .... ....

383 390 393 400 405 410 418 421 424 426 427 427

INTRODUCTION

Surface moving target indication (SMTI) involves searching Earth’s surface for moving objects using a dedicated radar mode. Ground-moving target indication (GMTI) is a subordinate, commonly referenced mode implying the detection, location, and discrimination of vehicles and personnel (dismounts) against rural, suburban, and urban land settings. The general class of SMTI radar modes also includes searching for vessels against sea, lake, and riverine backgrounds. An SMTI platform can be an aircraft, an unmanned aerial system (UAS), a satellite, an aerostat, or a tower. Typically, SMTI refers to object detection on Earth’s surface from airborne or spaceborne radar systems. In such aerospace radar configurations, the Doppler spread of mainbeam clutter is a fundamental limiting factor that impedes target detection and tends to drive system requirements and complexity. The use of side-looking, extended, along-track antennas; multichannel receive apertures; space–time adaptive processing (STAP); and low-pulse repetition frequency waveforms are common SMTI design choices to be described further in this chapter. SMTI interrogation strategies include wide area search (WAS); selective search, also sometimes called small area ground (SAG) mode; and persistent area search. Legacy system development focused on

383

384

CHAPTER

9

Surface Moving Target Indication

WAS-SMTI, but newer capabilities deployed on UASs and smaller platforms generally employ SAG or persistent search modes in addition to wider search capability. In a typical SMTI system, the primary challenge centers on mitigating the impact of mainlobe clutter on slow-moving-target detection. The returns from stationary clutter objects exhibit a distinct angle-Doppler region of support, since the specification of angle to a point on Earth’s surface also specifies its Doppler frequency shift. The objective of SMTI is to discriminate the target’s angle-Doppler response from that of the clutter background, assuming the target signal power is sufficiently strong relative to the receiver noise strength. SMTI has a number of military and civilian applications. Military applications include detection of troop movements, monitoring cordoned areas, and searching for time-critical targets in remote areas. Civilian applications include support for emergency management, such as evacuation of cities and towns, border surveillance, and highway safety. The Joint Surveillance Targeting Attack Radar System (Joint STARS) is an example of an airborne GMTI radar system; it carries the AN/APY-3/-7 military designation. The Joint STARS system, shown in Figure 9-1, houses a 7.3-meter-long, X-band antenna system mounted along the fore section of the fuselage in the readily identified, canoeshaped radome. The antenna is electronically scanned in azimuth and mechanically scanned in elevation. Three receive channels are adaptively combined to mitigate clutter and enhance detection performance. The Boeing 707 airframe supports long mission times, long-range search, and wide area coverage. The radar system outputs ‘‘dots’’ that represent the location and temporal appearance of detections. Figure 9-2 is an example of a GMTI display. The GMTI dots correspond to radar measurements of target position, range rate, signal-to-interference-plus-noise ratio (SINR), and other important attributes. An overview of Joint STARS and the field of GMTI radar is given in [1]. Another example is the AN/APY-9 UHF radar on the E2-D Advanced Hawkeye; this radar system incorporates a multichannel antenna and STAP to provide SMTI capability. Deploying GMTI on UAS platforms is one area of significant current interest. The Global Hawk UAS is an ideal GMTI platform; this system is shown in Figure 9-3. Global Hawk carries a substantial payload on the order of 900–1,400 kg (depending on model), flies at an altitude of 18 km at a speed of 160–175 m/s, and provides exceedingly long endurance [2]. The Multi-Platform Radar Technology Insertion Program

FIGURE 9-1 ¢ Joint STARS, AN/APY-3/-7 GMTI system [after http://www.af.mil/ information/ factsheets)].

9.1

Introduction

385 FIGURE 9-2 ¢ GMTI ‘‘Dots’’ shown in Local, Plan View Coordinates [after http://en.wikipedia. org/wiki/ Joint_STARS].

FIGURE 9-3 ¢ Global Hawk UAS Payloads Include Radar Capability [after http://www.af. mil/information/ factsheets)].

(MP-RTIP) is a scalable payload approach to GMTI and forms a basis for the Global Hawk GMTI mode. MP-RTIP became an option after publication of [1], hence its omission therein. The MP-RTIP architecture for Global Hawk employs multiple channels as required to produce an effective GMTI payload. The Vehicle and Dismount Exploitation Radar (VADER) developed by the Defense Advanced Research Projects Agency (DARPA) is designed for deployment on a Warrior UAS or similar platform. VADER provides the ability to ‘‘monitor a road, track a vehicle to a stop, observe dismount motion near the vehicle, characterize certain motions (like someone carrying a heavy load), and measure a ground disturbance after the vehicle departs’’ [3]. The AN/ APY-8 Lynx radar provides synthetic aperture radar (SAR) and GMTI capability in a

386 FIGURE 9-4 ¢ Canadian RADARSAT-2. [after http://www. radarsat2.info/ about/construction/ index.asp)].

CHAPTER

9

Surface Moving Target Indication Bus Module

Solar Panels Extendable Support Structure (ESS)

Solar Panels

SAR Antenna

lightweight package less than 37 kg. The AN/APY-8 provides GMTI detection range out to 23 km using a 44-cm by 16.5-cm Ku-band antenna at 320-W peak power [4]. The AN/ APY-8 is specially designed for UAS platforms. There has also been considerable interest in deploying SMTI capability from satellites. The space radar program included high-performance GMTI as a primary mission goal, with target characterization and persistent tracking of keen interest [5]. Typical satellite-borne GMTI modes are dependent on STAP to cope with limited alongtrack aperture relative to the satellite velocity (exceeding 7 km/s at low-Earth orbit, LEO). A STAP-based, spaceborne GMTI processing architecture is discussed in [6]. The Canadian RADARSAT-2 performs remote sensing using a C-band SAR; RADARSAT-2 builds on the legacy of the RADARSAT-1’s 15-m along-track aperture, push-broom SAR, with the addition of enhanced imaging modes and two along-track channels providing the necessary degrees of freedom to suppress clutter and detect moving targets [7] (see Figure 9-4). The RADARSAT-2 SMTI capability is called moving object detection experiment (MODEX) [8]. RADARSAT-2 uses, among other experimental approaches, a technique called along-track interferometry (ATI) to detect moving targets. In ATI, the processor compensates for the relative effects of motion between the two channels compared to the fixed scene; a conjugate multiply of one channel against the other zeros the phase response of the clutter background; a nonzero phase indicates the presence of moving targets [8]. Due to the importance of GMTI/SMTI, North Atlantic Treaty Organization (NATO) nations have agreed upon a standard GMTI format, called Standard NATO Agreement (STANAG) 4607. Details on STANAG 4607 are readily found in the public domain and available from NATO.

9.1.1 Organization This chapter aims to familiarize the reader with the tenets of SMTI radar. With this goal in mind, the remainder of this chapter is organized as follows. We describe SMTI radar operation in Section 9.2, including basic system issues, as well as target and clutter phenomenology that are driving system design and signalprocessing choices. We also consider SMTI radar search strategies. An understanding of surface clutter, target, and radio-frequency interference (RFI) signal characteristics is central to a discussion of SMTI radar. We describe multichannel, multipulse, SMTI signal models in Section 9.3. Measured data are included to validate the ground-clutter model. A primary objective of SMTI radar is to detect a target of a given radar cross section (RCS), with a specified minimum velocity, at a given range and with a desirable area coverage rate (ACR). Clutter, RFI, and receiver noise impede this process. The

9.1

Introduction

SMTI radar interrogates the scene and then must decide between two primary models that generated the observations in the radar receiver: (1) the null hypothesis (H0), or the case of target absence; or, (2) the alternative hypothesis (H1), which denotes target presence in addition to receiver noise, clutter returns, and RFI. Calculating and evaluating SMTI radar performance metrics is thus a critical undertaking. We discuss important SMTI metrics in Section 9.4. Exploiting the radar measurement space is the essence of radar detection. With the appropriate supporting hardware, the radar can collect measurements over fast-time, slow-time, spatial, polarimetric, and multiple-pass domains. Applying a matched filter to the fast-time samples yields range information, while the Fourier transform of slowtime samples yields Doppler. Each measurement accessible to the signal processor is called a degree of freedom (DoF). The SMTI processor combines those DoFs that best differentiate the target from clutter and RFI signals. Ground-clutter returns exhibit a well-defined angle-Doppler region of support that differs from the moving target response. The goal of SMTI is to discriminate the angle-Doppler offset of the target relative to the ground-clutter response. Section 9.5 describes SMTI antenna and waveform considerations. We consider a number of clutter-mitigation approaches in Section 9.6, starting with the one-dimensional, nonadaptive Doppler processing method, and then varying the DoFs and use of adaptive or deterministic signal processing to include discussion of adaptive Doppler processing, displaced phase-center antenna (DPCA) processing, adaptive DPCA, and STAP. Sections 9.7 and 9.8 describe the details of SMTI detection processing and target parameter estimation. In Section 9.7, we consider the practical matter of clutter filter normalization, constant false alarm rate (CFAR) processing, and postdetection integration. Section 9.8 develops a maximum likelihood approach to angle and Doppler estimation. Section 9.9 discusses additional considerations, including the impact of spatially varying, heterogeneous clutter on SMTI performance; detection of moving targets in SAR phase history data; bistatic or multistatic radar topologies; and dismount detection.

9.1.2 Key Points Key SMTI points the reader should identify in this chapter include the following: ●

SMTI radar attempts to detect moving vehicles, ships, boats, and dismounts (people) on Earth’s surface against strong clutter backgrounds. SMTI targets typically fall within the Doppler spread of mainlobe clutter and hence are referred to as being endoclutter. The along-track length of the antenna subsystem relative to the platform velocity is a critical factor determining the minimum detectable velocity (MDV) of the radar. Exploiting the radar’s angle and Doppler measurements are essential to effective SMTI radar performance. Deriving angle information requires a multichannel radar system. Discriminating slight differences in the target’s angle-Doppler response relative to that of the fixed clutter background is an important concept. Effective SMTI systems require sophisticated signal processing to achieve performance requirements. STAP is a preeminent approach to mitigate the effect of ground clutter and RFI signals on performance, thereby maximizing detection capability.

387

388

CHAPTER

9

Surface Moving Target Indication

STAP is related to the class of super-resolution methods and thus provides an MDV that a nonadaptive radar can only achieve by significantly increasing the along-track length of the radar antenna by a factor of 2 to 5. This is an important consideration when implementing SMTI from UAV or satellite-borne platforms. Challenges in SMTI include operation in complex, heterogeneous clutter environments and detecting and characterizing dismount targets. Advanced signal-processing methods are critical to achieve the best possible system performance.

9.1.3 Notation and Acronyms Basic Notation Frequently used variables are given below. f0 ¼ center frequency ðHzÞ; fd ¼ Doppler frequency ðHzÞ; f; q ¼ azimuth and elevation ðradsÞ; yg ¼ grazing angle ðradsÞ; l ¼ wavelength ðmÞ; M ¼ number of channels; N ¼ number of pulses; L ¼ number of available range bins; K ¼ number of range bins used in covariance estimate; T ¼ pulse repetition interval ðsÞ; c ¼ velocity of propagation ðspeed of light; m=sÞ; dr ¼ range resolution; ground plane ðmÞ; dcr ¼ cross-range resolution ðmÞ; ss ¼ spatial steering vector; vs ¼ hypothesized spatial steering vector; st ¼ temporal steering vector; vt ¼ hypothesized temporal steering vector; sst ¼ space-time steering vector; vst ¼ hypothesized space-time steering vector xs=k ðnÞ ¼ spatial data snapshot; kth range cell; nth pulse; xt=k ðmÞ ¼ temporal data snapshot; kth range cell; mth channel; xk ¼ space-time data snapshot; kth range cell; ck ¼ clutter space-time snapshot; kth range cell; j k ¼ RFI space-time snapshot; kth range cell; nk ¼ uncorrelated noise space-time snapshot; kth range cell; tk ¼ target space-time snapshot; kth range cell; Rk ¼ null-hypothesis covariance matrix; kth range cell; Rc=k ¼ clutter covariance matrix; RRFI ¼ RFI covariance matrix; b k ¼ null-hypothesis covariance estimate; kth range cell; R wk ¼ space-time weight vector; b k ¼ adaptive space-time weight vector; w yk ¼ filter output for kth range bin and selected angle and Doppler;

9.1

PD ¼ probability of detection; PFA ¼ probability of false alarm: Acronyms A/D ACR AMF CFAR CMT CNR CPI DoF DPCA FAD GMTI HRR ICM iid I/Q MDV MLE MVDR NCA PDI PRF PRI RCS RDM RFI RMS ROC Rx SAG SAR SINR SMI SMTI SNR STANAG STAP WAS

analog-to-digital (converter) area coverage rate adaptive matched filter constant false alarm rate covariance matrix taper clutter-to-noise ratio coherent processing interval degree of freedom displaced phase-center antenna (processing) false alarm density ground-moving target indication high-range resolution intrinsic clutter motion independent and identically distributed in-phase and quadrature (voltages) minimum detectable velocity maximum likelihood estimate minimum variance distortionless response noncoherent addition postdetection integration pulse repetition frequency pulse repetition interval radar cross section range-Doppler map radio frequency interference root mean square receiver operating characteristic receiver small area ground (search) synthetic aperture radar signal-to-interference plus noise ratio sample matrix inversion surface moving target indication signal-to-noise ratio Standard North Atlantic Treaty Organization Agreement space–time adaptive processing wide area search

Introduction

389

390

CHAPTER

9.2

Surface Moving Target Indication

9

SMTI RADAR OPERATION

The SMTI radar system searches an area on Earth’s surface, generally using a pulsed, linear frequency modulated (LFM) waveform. The coherent-processing interval (CPI) is comprised of the transmission and reception of N pulses at a common center frequency, f0. The pulse repetition interval (PRI) is T. A collection of M subapertures comprise the receive antenna. Each receiver channel down-converts and digitizes the receive waveform. The resulting data are conceptually organized into the radar data cube shown in Figure 9-5. To mitigate target fading effects – an aspect-dependent phenomenon resulting from the coherent summation of the many scatterers comprising the target – the radar frequency must hop several times within the dwell. Each hop corresponds to a CPI at the new center frequency. The radar generally will keep the waveform bandwidth constant. It is the waveform bandwidth that determines the radar-range resolution. The dwell time is given by NTP, where P is the number of CPIs in the dwell. Figure 9-6 shows the basic SMTI processing steps. The pulse compressor operates on each of the M channels, demodulating the waveform coding. Targets, clutter signals, receiver noise, and RFI may comprise the NM 1 space–time data vector corresponding to the kth range bin, xk. Receiver noise is temporally and spatially uncorrelated; it results

FIGURE 9-5 ¢ Coherent Radar Data Cube [after [25], † 2004 IEEE].

L Range Sample

Pre-Processing here RX

A/D

RX

A/D

RX

A/D

RX

A/D

RX

A/D

1

Receive Subarray

1

Fast Time

M

FIGURE 9-6 ¢ Generic SMTI Processing Architecture.

Target Angle and Doppler Estimates

Target Parameter Estimation

Pulse Compress

Multichannel Antenna

xk

Clutter/RFI Filter

Space–Time Data kth Range Bin

yk

Slow Time

N

Pulse

Postdetection Integration

Select CFAR Training Data

CFAR

Detections

1

9.2

SMTI Radar Operation

391

from thermal agitation of electrons within the receiver. Clutter signals and RFI, however, exhibit some degree of correlation and filtering mitigates their effects on detection performance. It is common to apply the linear filter, yk ¼ wH k xk , where wk are the filter weights. This linear filter formulation provides the framework needed to discuss Doppler processing, adaptive Doppler processing, DPCA, adaptive DPCA, STAP, and STAP variants, all methods used to suppress clutter. Adaptive digital beam forming also fits within this framework and is used to suppress RFI. We subsequently discuss these different methods in this chapter. The clutter/RFI filter step of Figure 9-6 is applied to each of the P CPIs. Generally, the processor selects a single receive direction and then generates the yk for all Doppler frequencies of interest. The result is a range-Doppler map (RDM). The processor then noncoherently sums the P RDMs for each of the CPIs in a step known as postdetection integration (PDI) to mitigate target fading. A detection threshold is then applied to the composite range-Doppler image. It is common to use a constant false alarm rate (CFAR) algorithm to set the detection threshold. CFAR algorithms estimate the varying residual interference power and scale the result in an attempt to keep the false alarm rate as constant as possible. After the processor detects a target, it then estimates bearing angle and Doppler frequency, sometimes also refining the range estimate. Bearing angle, Doppler, and range are then supplied to the analyst or an automatic tracker. The SMTI radar may switch to a different waveform to collect a high-range resolution (HRR) profile to facilitate target characterization and association. The HRR collection step is not shown in Figure 9-6. Figure 9-7 depicts several common SMTI collection modes. The wide area search mode involves searching large swaths of Earth’s surface using the bar-scanning strategy shown in the figure. Each beam position corresponds to a dwell, with P CPIs comprising each dwell as previously discussed. In the small area ground mode, the radar still searches a specified though smaller area. Usually in SAG mode, the radar increases the dwell time. SAG is used to look for more challenging target sets, like small vehicles. In the persistent search mode, the radar spotlights a designated area for an extended period. Usually, the radar uses significantly longer dwells in the persistent mode, generating

Persistent Search Mode

Down Range

Wide Area Search Mode Rtoe Bar 2 Rheel Rlength Rcenterf3dB

Small Area Ground Mode

Bar 1

Cross Range

FIGURE 9-7 SMTI Search Modes.

¢

392

CHAPTER

9

Surface Moving Target Indication

SAR and SMTI products concurrently. The WAS and SAG modes are very similar in design and are the underlying focus of the discussion in this chapter. The primary requirement of an SMTI radar is to detect a target of a specified RCS at a maximum range and minimum velocity while maintaining a desired area coverage rate. Unlike airborne moving target indication (AMTI) radar systems, SMTI radars usually do not operate in the noise-limited condition where receiver noise is the limiting factor. Rather, most SMTI radars are clutter limited. Developing a radar system meeting the SMTI requirements involves careful platform selection, antenna design, waveform selection, and signal-processing design and implementation. Key SMTI requirements are quantified using probability of detection (PD), false alarm density (FAD), minimum detectable velocity (MDV), area coverage rate, and bearing and Doppler root mean square (RMS) error. It is also common to characterize SMTI performance using SINR loss, a measure of the performance degradation relative to some other condition, usually the noise-limited capability or the difference between the optimal and adaptive filter. We consider SMTI metrics in detail in Section 9.4. As just mentioned, SMTI radar systems generally operate in the clutter-limited condition. This should not be a surprise, since the radar antenna is pointed toward Earth’s surface, generating strong surface reflections that mask the detection of moving targets of interest. Figure 9-8 depicts SMTI clutter-limited detection. To the left is shown angle-Doppler images for each range bin; clutter lies along a ridge, as given in the figure. The vertical boxes in two of the angle-Doppler images correspond to a particular transmit-and-receive angle. The Doppler spread across the main beam is DD. Antenna along-track length and platform velocity determine the magnitude of DD. Exoclutter FIGURE 9-8 ¢ Clutter-Limited Detection Performance.

Range Bin k Doppler

ΔD ΔD

T2

T1

Doppler

Range

Range Bin k-1

Doppler

Range T1 Bin k-2

T2 Mainlobe Clutter Angle

Doppler

9.3

Signal Models

target, T1, and endoclutter target, T2, are shown in range bin k – 2. Each vertical Doppler slice at the transmit–receive angle forms a horizontal slice at the corresponding range in the RDM to the right. Target T1 maps away from main beam clutter and is easily detected, whereas target T2 maps to the clutter edge. SMTI design – platform, aperture, signal-processing algorithms – focuses on improving the detectability of T2. As shown in Figure 9-8, targets T1 and T2 reside off of the clutter ridge, since their angle-Doppler coupling differs from that of the stationary clutter. A two-dimensional notch filter, with the null aligned along the clutter angle-Doppler region of support, is the ideal approach to enhance detection performance. Accessing angle information requires a multichannel antenna. The various antenna channels sample the direction of arrival of the propagating electromagnetic wave. The radar processor also requires access to the multipulse data to build the space-and-slow-time (angle-Doppler) filter. Note that clutter is generally white (uncorrelated) in range. Improved performance comes at the cost of additional receivers, more complex antenna array design, and more capable embedded computing systems. The following sections expand on the key elements of SMTI radar operation.

9.3

SIGNAL MODELS

We first discuss the space–time response of a point scatterer. The point-scatter model provides insight into the ground clutter, target, and RFI signal models. Consider the LFM transmit signal, bt ^2 stx ðt; nÞ ¼ a0 rect e j2pf0 t e jpgLFM t (9.1) tu where t is time, n is pulse number, a0 is the signal amplitude, bt ¼ t nT is fast time, T is the PRI, tu is uncompressed pulse width, f0 is the carrier frequency in Hertz, gLFM is the LFM chirp rate in Hz/s, and ( 1 rectðtÞ ¼ 1 for jtj 2 0 otherwise

(9.2)

Next, consider the case of transmit energy reflected from a target or source of point clutter. The received signal is a scaled, time-delayed replica of the transmit signal in (9.1), appearing as bt td j2pf ðtt Þ jpg ðbtt Þ2 sr ðt; nÞ ¼ a1 rect e 0 d e LFM d (9.3) tu

393

394

CHAPTER

9

Surface Moving Target Indication

with a1 a complex gain term proportional to the square root of the object’s RCS, and td the round-trip time delay from transmit phase center to the object and back to the receive antenna. Note that td typically varies with pulse number, n. After conversion to baseband and discrete-time sampling, (9.3) appears as ðkTs nTÞ td ðnÞ j2pfo td ðnÞ jpgLFM ðkTs nTtd ðnÞÞ2 e s r ðk; nÞ ¼ a1 rect e (9.4) tu k is the sample index and Ts is the sample rate. The options for conversion to complex baseband include the use of synchronous detection [separate in-phase and quadrature (I/Q) receive channels] or digital I/Q, the latter involving data oversampling, low-pass filtering, and decimation [9]. We explicitly show the dependence of td on the pulse index, n. Further, assume M receive channels comprise the receive aperture. The signals in each channel are generally similar, with a spatial time delay being the predominant difference as a result of the signal’s direction of arrival. The multichannel form of (9.4) is then ðkTs nTÞ td ðn; mÞ j2pfo td ðn;mÞ jpgLFM ðkTs nT td ðn;mÞÞ2 e e (9.5) ~s r ðk; n; mÞ ¼ a1 rect tu where m is the channel index. Examining (9.5), we observe a1 is the amplitude-modulation term resulting from the interaction of the waveform and scattering object; the first exponential term incorporates a phase modulation varying in space and time, 2pf0td(n, m); and we identify the quadratic, time-varying phase modulation corresponding to the LFM waveform in the argument of the latter exponential. [Equation (9.5) is easily generalized to other types of transmit waveforms. The LFM waveform is the most popular.] The linearly varying phase modulation, 2pf0td(n, m), encodes the object’s Doppler frequency and direction of arrival information. It is this space–time measurement the SMTI processor exploits to mitigate clutter and cohere the target’s return signal. The range resolution, determined by the bandwidth of the LFM waveform, is chosen to restrict the target to a single range bin over the coherent dwell. Pulse compression, applied on a pulse-by-pulse basis, removes the LFM modulation; the typical output of the pulse-compression stage appears as a weighted sinc function. Common range resolutions of tens of meters to single meter accuracy lead to sample intervals, Ts, on the order of tenths to hundredths of microseconds. The analog-to-digital converters in each receive channel provide this sampling mechanism. In contrast, the pulse-repetition interval is ordinarily on the order of milliseconds and is chosen to balance the trade-off between range and Doppler ambiguity. The time delay across elements of the multichannel array can be on the order of tenths to hundredths of nanoseconds for angles approaching end-fire. Pulse compressing (9.5), simplifying by grouping complex gain terms, and considering the peak range sample, kp, yields _ s r ðkp ; n; mÞ _

_

¼ a 1 ðkp Þej2pfo td ðn;mÞ

(9.6)

where a 1 ðkp Þ is a complex gain. A coherent space–time aperture for a typical system involves integrating tens to hundreds of pulses (e.g., 64 N 512) and usually several

9.3

Signal Models

to a dozen receive channels (2 M 12). Section 9.5 further considers coherent dwell length and design of spatial channels. It is common to describe the SMTI observation as a space–time data vector for the kth range bin of interest. The space–time data vector for the point scatterer follows from (9.6) as _ xk ¼ a 1 ej2pfo td ðn;mÞ m ¼ 1 : M n¼1:N

_

¼ a 1 ½ ej2pfo td ð1;1Þ

ej2pfo td ð1;2Þ

:::

ej2pfo td ðN ;M1Þ

T ej2pfo td ðN ;MÞ

(9.7)

where the superscript T denotes the transpose operation. Doppler frequency is given as the time derivative of –2pf0td(n, m), where m is fixed; normalizing the derivative by 1/2p yields Doppler in Hertz. Analogously, the spatial derivative of –2pf0td(n, m), for n fixed, yields spatial frequency, the indicator of signal direction of arrival. Equation (9.7) is expressible as _

_

xk ¼ a 1 ss ðf; qÞ st ðfd Þ ¼ a 1 sst ðf; q; fd Þ

(9.8)

with denoting the Kronecker product [10]; f is azimuth angle, q is elevation angle, fd is Doppler frequency, ss is the spatial steering vector describing the signal response across the receive array, st is the Doppler steering vector, ss–t is the space–time steering _ vector, and a 1 is the peak complex gain. The expression in (9.8) forms a basis for the following clutter, target, and RFI models.

9.3.1 Surface Clutter Surface clutter is the most significant impediment to SMTI detection performance. Ground clutter exhibits a distinct angle-Doppler coupling: small changes in angle result in small changes in Doppler, with knowledge of angle uniquely specifying Doppler frequency (assuming a smooth Earth). In a side-looking radar configuration, clutter lies along a ridge in the angle-Doppler plane; the ridge opens up into an ellipse when platform yaw is present and into a circle in the forward-looking collection geometry. The reflectivity of ground and sea clutter typically appear spatially varying. The motion of vegetation blowing in the wind or the rise and fall of the ocean’s surface leads to partial decorrelation of the clutter signal over the temporal aperture. The surface clutter space–time snapshot is the aggregate of the clutter returns from the various scatterers distributed in azimuth along the iso-range contour, including returns from ambiguous ranges. These scatterers include trees, grassy terrain, roadways, buildings, fence lines, etc. Dividing the azimuth interval into small angular extents representing a fraction of the lesser of the system spatial or Doppler resolution cells yields Nc clutter patches for each of Na range ambiguities. Each clutter patch incorporates a number of unresolvable, subscatterers whose composite return assures the received clutter voltage is random and independent among patches. The randomness of the composite return leads to clutter speckle, or fading: voltages from each of the

395

396

CHAPTER

9

Surface Moving Target Indication

subscatterers comprising the clutter patch add coherently to yield a constructive (strong) or destructive (weak) clutter voltage response. The surface clutter space–time snapshot takes the form ck ¼

Na X Nc X

ast ðm; n; kÞ sst ðfm;n ; qm;n ; fd=m;n ; kÞ

(9.9)

m¼0 n¼1

where ðfm;n ; qm;n Þ is the azimuth and elevation to the mnth patch; fd=m;n is the corresponding Doppler frequency; ast ðm; n; kÞ 2 C NMx1 is the vector containing the space–time voltages for each channel-pulse-range sample, with each element proportional to the square root of the patch clutter-to-noise ratio (CNR); sst ðfm;n ; qm;n ; fd=m;n ; kÞ is the space–time steering vector; and is the Hadamard product operation [10]. The space–time steering vector is given as the Kronecker product of the temporal and spatial steering vectors, sst ðf; q; fd Þ ¼ st ðfd Þ ss ðf; qÞ. The voltage vector can be written as (9.10) ast ðm; n; kÞ ¼ vk=m;n d t ðfm;n ; qm;n ; kÞ d s ðfm;n ; qm;n ; kÞ where vk=m;n is the complex voltage of the mnth patch for the kth range, d t ðfm;n ; qm;n ; kÞ is a random taper vector characterizing the voltage fluctuation over the temporal aperture, and d s ðfm;n ; qm;n ; kÞ is a random spatial taper describing the voltage decorrelation over the spatial aperture. We recognize both d t and d s as covariance matrix taper (CMT) components [11]. Let vk=m;n be zero mean, complex normal, with variance s2k=m;n , i.e., vk=m;n CNð0; s2k=m;n Þ. The covariance matrix of (9.10) then follows as 2 Dk ðm; nÞ ¼ E½ast ðm; n; kÞaH st ðm; n; kÞ ¼ sk=m;n Dt ðfm;n ; qm;n ; kÞ Ds ðfm;n ; qm;n ; kÞ; Dt ðfm;n ; qm;n ; kÞ ¼ E½d t ðfm;n ; qm;n ; kÞd H t ðfm;n ; qm;n ; kÞ; Ds ðfm;n ; qm;n ; kÞ ¼ E½d s ðfm;n ; qm;n ; kÞd H s ðfm;n ; qm;n ; kÞ:

(9.11) The expected value of the two outer products corresponds to temporal and spatial CMTs. Also, the superscript, H, denotes conjugate transpose. Plausible functions characterizing the elements of Dt, suitable for modeling intrinsic clutter motion, include the Gaussian autocorrelation [12] and Billingsley model involving an exponential autocorrelation [13]. Gaussian fits best for regions with seawater or freshwater because it fully decorrelates, whereas exponential is more appropriate for wooded regions or fields. A sampled sinc or ‘‘angle dither’’ is appropriate for the elements of Ds and used to model wavefront dispersion [11].

Signal Models

9.3

397

The clutter covariance matrix, Rc=k ¼ E½ck cH k , follows from the prior discussion as Rc=k ¼

Na X Nc X

s2k=m;n Dst ðfm;n ; qm;n ; kÞ sst ðfm;n ; qm;n ; kÞsH st ðfm;n ; qm;n ; kÞ;

m¼0 n¼1

(9.12)

Dst ðfm;n ; qm;n ; kÞ ¼ Dt ðfm;n ; qm;n ; kÞ Ds ðfm;n ; qm;n ; kÞ:

The clutter covariance matrix is unknown in practice. Many SMTI signal-processing architectures estimate the unknown clutter-plus-noise covariance matrix using training data. Observe that (9.12) is a function of range bin, k, thus acknowledging the potential for spatially varying clutter properties. As discussed in [14], the model in (9.12) provides a good match to measured data. Figure 9-9 shows minimum variance distortionless response (MVDR) spectra (superresolution views, see [10]) of the clutter angle-Doppler behavior for measured data taken from the Multi-Channel Airborne Radar Measurements (MCARM) program; the spectra in Figure 9-9 use estimated covariance matrices that average data over the range bins identified in the figure. Figure 9-10 shows the simulated response using the model of (9.12) and the radar parameters corresponding to the CPI shown in Figure 9-9.

MCARM MVDR Spectrum, Near Range

MCARM MVDR Spectrum, Midrange

0 600 200

–10

600

–15

400

–20

–25

–200

–30

–400

–40

–800

60

90

–15 –20

–400

–800 –30 0 30 Angle (Degrees)

–10

–600

–60

–5

–200

–35

–45

–90

–25

–60

–30 0 30 Angle (Degrees)

60

90

MVDR Spectrum for Simulated MCARM Data

800

–5

600

–10

400

–15

200

–20

0 –25

–200

–30

–400 –600

–35

–800

–40

–90

–60

–30

30

Angle (Degrees)

60

90

FIGURE 9-9 ¢ Measured MCARM MVDR Spectra with Training Data Selected Over (a) Bins 200 to 300 and (b) Bins 350 to 450 [after [14], † 2006 IEEE].

–30

FIGURE 9-10 ¢ MVDR Spectrum for MCARM Simulation [after [14], † 2006 IEEE].

dB

–90

Train Bins [350:450]

200

–600

Doppler (Hz)

Doppler (Hz)

400

–5

800

dB

Train Bins [200:300]

dB Doppler (Hz)

800

398

CHAPTER

9

Surface Moving Target Indication

Comparing Figure 9-10 with Figure 9-9, we see a very good match between the clutter responses. The primary difference is the end-fire clutter power spread observed in the measured data results due to multipath from the wing and radome struts corresponding to the port-mounted antenna system. Equation (9.12) does not model these multipath effects, which appear very near the system noise floor and have virtually no bearing on detection performance. Moreover, typical field deployments mount the antenna system under the fuselage to avoid such anomalous characteristics.

9.3.2 Targets A description of different SMTI target models is given in Table 9-1. It is commonplace to use the fluctuating Swerling models to characterize SMTI targets [12]. The definitions in Table 9-1 differ slightly from [12] to account for modern radar operation. In the Swerling 1 model, the target voltage is circular Gaussian as a result of the interaction of similarly sized subscatterers adding coherently and yielding, for certain aspects, a strong return, while for others a weak response. Otherwise, the target voltage is assumed perfectly correlated over the CPI. The weak response is called target fading. A Swerling 2 target exhibits decorrelated voltage responses from sample to sample. In the Swerling 2 case, the samples are target pixels taken from the RDMs generated for each of P CPIs, with each CPI corresponding to an offset and nonoverlapping transmit frequency. Frequency hopping purposely leads to target decorrelation and overcomes target fading effects. To avoid the deleterious impact of fading, the SMTI radar employs offset center frequencies for each of the CPIs comprising the dwell. For example, an X-band SMTI radar might combine five 100-millisecond CPIs using a 30-MHz LFM waveform at center frequencies of 10 GHz, 10.03 GHz, 10.06 GHz, 10.09 GHz, and 10.12 GHz. The processor outputs from each CPI are then noncoherently combined to enhance detection performance. The envelope probability density function (PDF) is chi-squared for Swerling 3 and Swerling 4, with correlation properties matching Swerling 1 and Swerling 2; the presence of a dominant scatterer leads to the chi-squared envelope. It is rare to assume the target is nonfluctuating. In the nonfluctuating model, the SNR is fixed, since the RCS is constant for all aspects. Sometimes the nonfluctuating target model is used in simulations to test and evaluate signal-processing algorithms.

TABLE 9-1

¢

SMTI Target Models

Target Type

Complex Envelope PDF

Swerling 1

Rayleigh

Swerling 2

Rayleigh

Swerling 3 Swerling 4 Nonfluctuating (‘‘Swerling 0’’)

Chi-squared Chi-squared Delta function

Temporal Correlation Perfectly correlated over the CPI, RCS fluctuates due to changes in target aspect from CPI to CPI. Uncorrelated from CPI to CPI, RCS fluctuates mainly due to frequency hopping between CPIs. Same as Swerling 1. Same as Swerling 2. Perfectly correlated over CPI, RCS is a constant value, yielding fixed SNR.

9.3

Signal Models

Most radar-system analyses exclusively rely on the Swerling 1–Swerling 2 model set. The target snapshot for Swerling 1 and Swerling 2 targets is tk ¼ at sst ðf; q; fd Þ

(9.13)

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi where at ¼ SNR Pn V; Pn is the uncorrelated, receiver noise power; and V is a zero mean, unity variance complex Gaussian variate, viz. V ~ CN(0,1). With each frequency hop, V is redrawn to model the Swerling 1–Swerling 2 model.

9.3.3 Interference RFI generally results from the reception of an in-band waveform generated by another electronic device, such as a communication system. The RFI waveform is uncorrelated with the radar waveform. The RFI data snapshot is xJ ðnTÞ ¼ wðnTÞ ss ðf; qÞ

(9.14)

The RFI waveform, w(t), exhibits noiselike characteristics, abiding for discrete times nT and qT by E½wðnTÞw ðqTÞ ¼ s2J dððn qÞT Þ

(9.15)

where d(nT) is the digital delta function and s2J is the RFI power. In this model, the RFI exhibits spatial correlation, as it emanates from a specific angle relative to the radar, but the RFI waveform is white over the temporal aperture. The resulting space–time covariance matrix is RRFI ¼ I N s2J ss ðf; qÞsH (9.16) s ðf; qÞ In (9.16), IN is the N N identity matrix. The space–time covariance matrix of the RFI is block diagonal as a result of (9.15); each block corresponds to a spatial covariance matrix identified within the parentheses of (9.16) that characterizes the RFI direction of arrival and power.

9.3.4 Space–Time Snapshot Two models characterize the SMTI radar space–time snapshot. Under the null hypothesis, the snapshot for range bin k is H0 : xst=k ¼ ck þ j k þ nk

(9.17)

whereas under the alternative hypothesis of target presence the snapshot appears as H1 : xst=k ¼ tk þ ck þ j k þ nk

(9.18)

nk is the uncorrelated receiver noise component, nk CN ð0; s2n INM Þ, where s2n is the receiver noise power, INM is the NM by NM identity matrix, and 0 is the zero vector (of length NM by 1, in this case). The snapshot is organized by stacking each spatial snapshot for a given pulse one on top of the other.

399

400

CHAPTER

Surface Moving Target Indication

9

Since clutter, interference, and noise signals are independent, the null-hypothesis covariance matrix is Rk ¼ Rc=k þ RRFI þ s2n INM

(9.19)

The SMTI radar signal processor manipulates the space–time snapshot at each range to determine which of the two hypotheses is most likely valid. Each of the elements of the space–time snapshot are coherently combined to ideally provide acceptable probability of detection at a required false alarm density.

9.4

SMTI METRICS

9.4.1 Probability of Detection The objective of the SMTI radar is to maximize probability of detection while maintaining an acceptable false alarm rate. The receiver operating characteristic (ROC) describes detection performance as a function of the probability of false alarm (PFA) and output SINR. Figure 9-11 shows the ROC curves for nonfluctuating (denoted Swerling 0) and Swerling 1 targets in circular Gaussian disturbance. SINR is shown using the target RMS power. (A ROC curve more commonly plots probability of detection versus probability of false alarm for varying SINR.) The difference between the Swerling 1 curve relative to the nonfluctuating case for a fixed probability of detection and false alarm rate is known as fluctuation loss. Figure 9-11 reveals that probability of detection increases monotonically with output SINR for a fixed probability of false alarm [15]. In addition, the figure shows that small changes in output SINR generally lead to large changes in probability of detection. Thus, it is critical the radar signal processor maximize output SINR to achieve best performance. Over the CPI, the target RCS is assumed constant. The constant RCS value, however, is a random variable with a Rayleigh-distributed complex envelope (or exponentially distributed power). This randomness describes the Swerling 1 fluctuating target model. According to the curve in Figure 9-11, the fluctuation loss, Lfluctuate, between the nonfluctuating (Swerling 0) target and the Swerling 1 target is about 8 dB at PD ¼ 0.9 Receiver Operating Characteristic

FIGURE 9-11 ¢ Receiver Operating Characteristic (ROC) Curve. Probability of Detection

1

0.8

0.6

0.4

PFA = 1e-06, Swerling 0 PFA = 1e-04, Swerling 0

0.2

PFA = 1e-06, Swerling 1 PFA = 1e-04, Swerling 1

0 –10

10 SINR (dB)

20

30

SMTI Metrics

9.4 TABLE 9-2

¢

Postdetection Integration Performance Gains, PFA ¼ 1E 6 PD ¼ 0:9, Lfluctuate ¼ 8 dB

I(n) (from NCAs [12])

Required Input SINR Fluctuating (dB)

Required Input SINR Nonfluctuating (dB)

1 2 3 4 5

21.2 13.7 11.2 9.7 8.2

13.2 10.2 8.4 7.2 6.2

0 7.5 10 11.5 13

PD ¼ 0:5, Lfluctuate ¼ 1:25 dB

I(n) (from [12]) (dB)

Required Input SINR Fluctuating (dB)

Required Input SINR Nonfluctuating (dB)

0 4 5.5 6.5 7.5

12.75 8.75 7.25 6.25 5.25

11.5 8.5 6.7 5.5 4.5

and PFA ¼ 1E – 6 and 1.25 dB at PD ¼ 0.5 for the same false alarm rate. To enhance performance against target fluctuation (or, equivalently, target fading), the SMTI radar system commonly frequency hops to decorrelate the target voltage from CPI to CPI, thus inducing Swerling 2 target behavior. Through noncoherent addition (NCA) of the voltage outputs from CPI-level processing, the radar achieves integration gain. This latter step is called postdetection integration. Table 9-2 describes the detection performance gain resulting from PDI. The first column gives the number of CPIs noncoherently summed. The table is then broken into halves: desired PD ¼ 0.9 and desired PD ¼ 0.5. The false alarm rate is PFA ¼ 1E – 6. IðnÞ is the integration gain resulting from the NCA of the uncorrelated, Swerling 2 target voltages relative to the single CPI case; these values are taken from [12]. The column labeled ‘‘Required input SINR fluctuating’’ is the required single CPI SINR input to the PDI process to achieve the desired detection probability. The column labeled ‘‘Required input SINR nonfluctuating’’ is the single CPI SINR required when the coherent integration time is increased by a factor equal to the number of NCAs and the target RCS does not fluctuate (Swerling 0). The required input SINR is always lower for the nonfluctuating case since there is no fading loss and the coherent integration is optimal. The required input SINR for the fluctuating case is given as the required Swerling 1 SINR from Figure 9-11 (the value when the number of NCAs is unity) minus the integration gain, I(n). The ‘‘Required input SINR fluctuating’’ is the requirement at the output of all coherent-processing steps, prior to PDI, to achieve the desired detection probability. As seen from Table 9-2, there is only slight performance loss between the fluctuating and nonfluctuating cases for PD ¼ 0.5. However, PDI provides substantial gain at the more useful, higher detection probabilities, with only a few decibels of SINR separating the performance between coherently integrated nonfluctating and PDI fluctuating cases.

9.4.2 False Alarm Density In SMTI radar, it is desirable to specify the number of false alarms per area. This false alarm density is calculated by converting the probability of false alarm to number of false reports per area under surveillance, Number of Decisions (9.20) FAD ¼ PFA Beam Area

401

402

CHAPTER

9

Surface Moving Target Indication

The beam area is approximately given by p Beam Area ffi ðRtoe Rheel Þsecytoe ðRcenter f3dB Þ 4

(9.21)

where Rtoe and Rheel are slant ranges corresponding to the 3-dB points of the antenna elevation beam pattern intersecting Earth’s surface, ytoe is the grazing angle at the toe, Rcenter is the slant range to scene center, and f3dB is the antenna’s 3-dB beamwidth in azimuth (see Figure 9-7). A suitable SMTI false alarm density is on the order of 0.1/km2. A high FAD masks targets of interest and makes target associations very difficult.

9.4.3 Signal-to-Interference-Plus-Noise Ratio Loss Signal-to-interference plus noise ratio (SINR) directly affects probability of detection, as discussed in Section 9.4.1. The output SINR of the linear filter, y k ¼ wH k xst=k ¼

NM X ½wk m ½xst=k m

(9.22)

m¼1

where wk 2 C NMx1 is the complex weight vector and denotes the conjugation operation, is E½yk=t y k=t Target Signal Power SINR ¼ ¼ Interference þ Noise Power E½yk=H0 y k=H0 ¼

2 H s2t jwH E½wH k sst ðft ; qt ; fd=t Þj k tk tk wk ¼ H E½wH wH k xk=H 0 xk=H 0 wk k Rk=H 0 wk

(9.23)

yk=t and yk=H0 are filter outputs for cases of target-only and the null hypothesis; ft , qt, and fd=t are target azimuth, elevation, and Doppler frequency; s2t is the target power; and, xk=H0 and Rk=H0 are the null-hypothesis snapshot and covariance matrix, respectively. SINR loss compares the performance degradation between a given processor implementation and an ideal. The most commonly used SINR loss term is defined as ! 2 H jw s ðf ; q ; f Þj st t d=t t k s2t 2 wH SINR s2 jwH sst ðft ; qt ; fd=t Þj k Rk=H 0 wk 2 (9.24) ¼ n k H Ls;1 ¼ ¼ NM SNR wk Rk=H 0 wk st NM s2n Equation (9.24) is sometimes called clairvoyant SINR loss, since its calculation requires knowledge of the null-hypothesis covariance matrix; it must be approximated when using real-world data. Factors influencing the SINR loss in (9.24) include the

SMTI Metrics

9.4

severity of the clutter and interference environment, the system design, and the particular algorithm defined by the selection of wk. Another commonly used SINR loss metric is called adaptive SINR loss. It compares the difference in performance between an adaptive filter and its optimal implementation. The adaptive filter must estimate unknown quantities in its implementation, viz., the null-hypothesis covariance matrix and the target space–time steering vector, and the adaptive SINR loss captures the impact of estimation error. Adaptive SINR loss is given by ! 2 bH jw k sst ðft ; qt ; fd=t Þj Ls;2 ¼

Adaptive SINR ¼ Clairvoyant SINR

bH bk w k Rk=H 0 w 2 jwH k sst ðft ; qt ; fd=t Þj wH k Rk=H 0 wk

!

(9.25)

b k is the adaptive weight vector, an estimate for wk computed using known quantities. w b k in its implementation. Ls;2 is a random variable and is For example, STAP employs w b k ¼ wk , it is seen Ls;2 ¼ 1 in (9.25). described in further detail in [16]. In the case where w In [16], it is shown that when the training data used to estimate the covariance b k (to be described subsequently) are independent and identically distributed matrix in w (iid) and the data are complex Gaussian, then the adaptive losses are roughly 3 dB on average when the processor uses a training data interval of roughly twice the processor’s degrees of freedom (e.g., NM for STAP). This is sometimes called the Reed-MallettBrennan rule after the authors of [16]. The SINR loss terms are all bound between zero and unity. Since the SINR loss terms are all applied directly as numerator terms to the radar-range equation, negative-valued losses in decibels are indeed losses. This might seem like an oxymoron, but it is standard convention in the SMTI and STAP communities. An example of SINR loss, Ls,1, is shown in Figure 9-12. This figure compares estimated SINR loss using data taken from the MCARM system against simulated loss using the clutter model from the prior section and the calculation in (9.24). The numbers in the legend correspond to the range bins used to estimate the clutter-plus-noise covariance matrix. While the MCARM system used a side-looking antenna, the clutter null in Figure 9-12 is shifted from 0 Hz due to a few degrees of yaw. The match between the simulated and measured data results is very good.

9.4.4 Minimum Detectable Velocity The minimum detectable velocity is a critical SMTI performance metric that drives aperture, platform, and algorithm selection. Minimum detectable velocity is defined as the minimum target velocity where the radar system provides an acceptable, specified probability of detection at the desired false alarm rate. The severity of the clutter environment affects MDV. ICM is a fundamental, limiting factor on radar MDV. MDV is approximated as follows: ●

Calculate, or estimate, the target SNR;

calculate, or estimate, the aggregate SINR loss, e.g., Ls,1 Ls,2;

403

404

CHAPTER

Surface Moving Target Indication

9

Estimated SINR Loss, Boresight Direction

FIGURE 9-12 ¢ Example SINR loss, measured and simulated data [after [14], † 2006 IEEE].

Estimated SINR Loss (dB)

−5 −10 −15 −20 400:500 500:630 Simulated

−25 −30 −35

−800 −600 −400 −200

200

400

600

800

Doppler (Hz)

● ●

estimate SNR enhancement resulting from noncoherent addition; and, finally, determine the lowest target velocity where the product of SNR, SINR loss, and noncoherent gain exceeds the SINR required to achieve the desired probability of detection for the specified probability of false alarm or FAD.

Usually, high-fidelity numerical simulation is used to calculate Ls,1 Ls,2. Estimating Ls,1 Ls,2 requires a good estimate of the radar noise floor and an excellent estimate of the array response relative to the ideal steering manifold.

9.4.5 Area Coverage Rate A common SMTI specification is: cover a specified area in a minimum allowable time, while providing a desired MDV against a certain target class at a maximum range designation. Figure 9-7 depicts the radar search in WAS mode. The beam area for a chosen bar is specified by (9.21). The total time in each beam is called the dwell time and is the sum of the CPI lengths corresponding to each fixed beam position. Factors influencing the CPI length include the desired integration gain, target acceleration, and required Doppler resolution. Then the area coverage rate is given in straightforward fashion by ACR ¼ ðArea=BeamÞ ðDwell=BeamÞ (9.26) The ACR is spatially varying. In this case, it makes sense to specify an average value. Equation (9.26) must consider beam overlap to provide the most accurate estimate.

9.5

Antenna and Waveform Considerations

9.4.6 Parameter Estimation Accuracy After detecting a target, the radar system generally attempts to estimate target location and speed. The radar first determines the target bearing (in antenna coordinates) and then, using knowledge of the platform location and antenna state, the radar processor maps the bearing estimate to a location within a fixed coordinate system. This coordinate system may be local – such as north–east–down (NED) – but given knowledge of the origin and orientation, is easily relatable to an Earth-centered framework, such as latitude–longitude–altitude (LLA) with World Geodetic System (WGS) standards. WGS-84 is the current standard. The processor further estimates the target range–rate through the Doppler frequency estimate and inertial navigation unit (INU) data coupled to a Global Positioning System (GPS) receiver. The INU/GPS provides estimates of the platform-velocity vector and antenna-pointing direction; this information is needed to map the measured Doppler frequency shift, comprising both target and platform motion, to the target range–rate. It is common to assess position and range–rate accuracy using root mean square error estimates. The RMS error is the square root of the square of the difference between the components of the actual parameter value and the estimate. When characterizing position error, it is common to report the total error. It is also common to specify the elliptical error probable (EEP) for 50-percent or 90-percent levels. EEP-50 is the ellipse projected onto Earth’s surface characterizing the accuracy of half the measurements. For example, a 1-m by 1-m EEP-50 indicates that 50 percent of the estimates are within a circle at least 1 m from the true location. (When the ellipse defaults to a circle, this measure is called the circular error probable, or CEP.) Similarly, EEP-90 indicates 90 percent of the targets fall within an ellipse on Earth’s surface with specified semimajor and semiminor axes.

9.5 ANTENNA AND WAVEFORM CONSIDERATIONS This section briefly overviews antenna and waveform issues important in SMTI radar operation.

9.5.1 Antenna The along-track length (horizontal array dimension), height, frequency, number of channels, and subarray design are important SMTI antenna design choices.

9.5.1.1 Along-Track Antenna Length The along-track antenna length determines the azimuth beamwidth. Coupled with the platform velocity, the azimuth beamwidth determines the spread of mainlobe clutter. Assume a side-looking array configuration, since the Doppler spread is greatest in this case. Let fc be the cone angle from the platform center line – and, hence, platform-velocity

405

406

CHAPTER

9

Surface Moving Target Indication

vector – to a particular point of interest. The Doppler frequency at the antenna 3-dB point closest to the velocity vector, and on Earth’s surface, is fd=1 ¼

2vp cosðfc DB Þ l

(9.27)

where vp is the platform velocity, l is wavelength, and 2DB is the 3-dB antenna azimuth beamwidth. The corresponding Doppler at the 3-dB point farthest from the velocity vector is fd=2 ¼

2vp cosðfc þ DB Þ l

(9.28)

The total mainbeam Doppler spread is the difference between (9.27) and (9.28); this spread has a profound impact on MDV. Usually, techniques like STAP can reduce the MDV from the diffraction-limited spread given by the difference between (9.27) and (9.28) by a factor of 2 or more. For nonadaptive methods, the radar dwell must increase substantially to reduce the MDV. This latter approach is not as efficient as increasing the along-track antenna length.

9.5.1.2 Antenna Height In SMTI radar, antenna height determines the elevation beamwidth and, hence, the footprint length (in range) on Earth’s surface. The footprint size impacts the highest selectable pulse repetition frequency (PRF) if the radar is to avoid range ambiguities within the antenna mainlobe. It is desirable to ensure that only one pulse traverses the footprint at any given time, since clutter fold-over increases the CNR and can appear spectrally divergent when the antenna normal and velocity vector are nonorthogonal. Given the platform height, hp, and the effective Earth radius, REM, both in meters, the relationship between the angle from nadir, qnadir, and the slant range, Rslant, is rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi Rslant ¼ ðREM þ hp Þcosqnadir

ðREM þ hp Þcosqnadir

2

hp ð2REM þ hp Þ

(9.29)

To determine the range to the antenna footprint toe and heel, take the antenna elevation beam-pointing angle and the antenna elevation beamwidth, convert the antenna elevation 3 dB points into nadir angles, and then substitute into (9.29). Next, the grazing angle, yg, follows as 2 Rslant hp ð2REM þ hp Þ yg ¼ arcsin (9.30) 2REM Rslant As (9.21) shows, the footprint length is (Rtoe – Rheel) sec ytoe. The unambiguous ground range is given by Ramb

cT secytoe 2

(9.31)

with c being the speed of light. To avoid ambiguities within the footprint, we require Ramb ðRtoe Rheel Þsecytoe

(9.32)

9.5

Antenna and Waveform Considerations

The PRF is then constrained as PRF

c 2ðRtoe Rheel Þ sec ytoe

(9.33)

Increasing the height of the antenna decreases the 3-dB elevation beamwidth and the length of the footprint, thereby allowing radar operation at a higher PRF. The higher PRF is useful to mitigate Doppler ambiguity and reduce the number of receive channels.

9.5.1.3 Subarray Design Considerations There are several approaches to design a multichannel radar: form subapertures from the elements of a phased array antenna, employ reflector antennas offset in elevation baseline and mechanically squinted in azimuth, or use horn antennas squinted to form multiple beams in azimuth and elevation. Forming multiple subapertures from a phased array antenna is generally the preferred approach, since the phased array radar most readily supports wide area search and other radar modes. A digital subarray is usually a collection of radiating elements, electronically combined to form a receive beam in a specified direction, and including a receiver and analog-to-digital converter. The SMTI radar will use M subarrays, or channels, to spatially sample propagating electromagnetic waves impinging on the array. In some cases, especially for lower frequency applications, the system combines a column of radiating elements into a single along-track channel. Figure 9-13 shows these example subarray configurations for M ¼ 4. Grating lobes (or high-peak sidelobes) are spatial ambiguities occurring for uniformly sampled arrays when the separation between spatial sample points exceeds l/2. Specifically, grating lobes occur in a uniform linear array at cone angles nl fc ¼ arccos cosfscan þ (9.34) d where fscan is the scan cone angle, d is the subarray channel spacing, and n is an integer chosen to yield a real solution, thereby satisfying 1 cosfscan þ nl d 1. (Note: A ULA measures cone angle, and this cone angle generally differs from the Doppler cone angle previously defined, depending on the antenna mounting. The Doppler and antenna cone angles are the same when the ULA is oriented along the platform-velocity vector. The difference between the two cone angles is evident from context in our discussion.) Grating lobes are problematic in SMTI for two main reasons. First, if beam spoiling is used on transmit, then care should be taken to ensure the mainlobe transmit pattern does not extend into the angle ambiguous receive region; otherwise, the processor suffers significant clutter loss in the look direction – but at Doppler frequencies away from the unambiguous clutter ridge extent. The appearance of additional clutter ridges intersecting the look direction lead to targetblind Dopplers. A secondary concern is the presence of RFI in the vicinity of a spatial ambiguity. Adaptive spatial nulling suppresses the RFI source at the expense of a distorted gain pattern in the desired look direction and higher sidelobes.

407

Surface Moving Target Indication

9

Subarray 1

X X X X

X X X X

X X X X

Subarray 2

X X X X

X X X X

X X X X

X X X X

λ 2

Subarray 3

X X X X

X X X X

X X X X

X X X X

Subarray 4

X X X X

X X X X

X X X X

X X X X

X X X X

X = Antenna Element

PA Phase Shifter

Antenna Element

LNA

(a)

PA Phase Shifter

To Transmit Waveform Generator or Digital Receiver

FIGURE 9-13 ¢ Examples of Subarray Configurations: (a) Nonoverlapped, 4 4 Subarray Configuration, with Each Antenna Element Tied to its Own Power Amplifier (PA), Low Noise Amplifier (LNA), Switches, and Phase Shifter; (b) a Column of Antenna Elements Called a Stick Summed Together and Fed to a Transmit–Receive Module.

CHAPTER

To Transmit Waveform Generator or Digital Receiver

408

Subarray 1 λ 2

LNA

PA Phase Shifter

Subarray 2

LNA

X X X X X

X X X X X

X X X X X

X X X X X

PA

Subarray 3

Phase Shifter

LNA

PA

Subarray 4

Phase Shifter

LNA

(b)

Overlapped subarrays represent another design option. In the overlapped subarray design, antenna elements are shared among receive channels. In this approach, phase centers are moved closer, thereby pushing grating lobes out toward the invisible region (beyond 90 ). Sharing subarray elements results in the correlation of receiver noise. The impact of element reuse and receiver noise correlation is generally minor. As shown in [17], M spatial channels support M – 1 nulls. As a general rule, an SMTI radar system requires at least three channels (M ¼ 3) to simultaneously null clutter and form a beam in the target search direction. This is the case for the Joint STARS radar, as mentioned in Section 9.1. Referring to Figures 9-8 through 9-10, when the clutter ridge is unambiguous in Doppler within the mainlobe, a single spatial null is

Antenna and Waveform Considerations

9.5

d

Doppler Process

Channel 3

Doppler Process

w*k/1,1

Σ

FIGURE 9-14 ¢ Three-channel SMTI Processing Architecture.

d Channel 2

Channel 1

w*k/1,2

Doppler Process

w*k/2,1

Σ

w*k/2,2

Compare Phase

>
H1 H1 jvH R k b1 xk j2 > T1 ðvH R b1 R ¼ st 1 T1 ) jvH st st k vst Þ k < < |fflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflffl} H0 H0 b v vH R st

k

(9.57)

st

Estimate of Interference Power

This expression applies to the post-Doppler case with the appropriate substitutions, as described in Section 9.6.3. To the right of the arrow, it is seen the AMF decision simplifies to the magnitude squared of the adaptive filter output compared to the equivalent of a threshold multiplier, T1, times an estimate of the local clutter power residue (to see this, consider (9.56) with b m st ¼ 1, rather than the AMF normalization). The AMF normalization is imperfect in heterogeneous environments, so other CFAR methods, like cell averaging CFAR, are used in practice [12]. The processor applies the aforementioned step to each of P CPIs, generating the RDMs for a common look direction. The transmit frequency hops for each CPI to mitigate target fading. It is possible to apply a detection threshold to each RDM and then implement M-of-N detection [12]. However, a preferable approach is to noncoherently add the RDMs in a step called postdetection integration. Section 9.4.1 describes the benefits of PDI. The processor applies a detection threshold to the noncoherently summed RDMs to determine target presence, as Figure 9-22 suggests. Either a fixed or CFAR threshold can be used; the fixed threshold tends, naturally, to suffer from a higher FAD. After a threshold crossing, the processor fetches the corresponding I/Q data and estimates the target bearing and Doppler. Details of angle-Doppler estimation are given subsequently in Section 9.8. The target range, Doppler, angle, and other characteristics – such as estimated SINR – are formatted and provided to postdetection processing. The STANAG 4607 is the common approach to format GMTI/SMTI data. In some cases, additional processing is used to cope with the impact of heterogeneous clutter on detection performance. The details of this sophisticated processing are beyond the scope of this chapter. A brief overview is given in Section 9.9.1 for the interested reader.

9.8

ANGLE AND DOPPLER ESTIMATION

Once the processor detects a target, a bearing estimate is necessary to initiate target tracking. A method seamlessly integrating with STAP processing is desirable [33–35].

421

422

CHAPTER

9

Surface Moving Target Indication

Express the alternative hypothesis of (9.18) as 2

3 f xk ¼ asst ðpÞ þ ntot ; p ¼ 4 q 5 fd

(9.58)

where a is a complex constant, and ntot ¼ ck þ jk þ nk is the interference-plus-noise (total noise) vector. The basic problem is to estimate the true target parameters, p, with pb. If ntot CNð0; Rk Þ, then the PDF is 1 pðxk jpÞ ¼ M exp ðxk asst ðpÞÞH R1 ðx as ðpÞÞ (9.59) k st k p jRk j Given (9.59) as a likelihood function, we first require an estimate for the complex constant. Equation (9.59) is maximal when Qða; pÞ ¼ ðxk asst ðpÞÞH R1 k ðxk asst ðpÞÞ

(9.60)

is minimal. Differentiating (9.60) and setting the result to zero yields b¼ a

1 sH st ðpÞRk xk 1 sH st ðpÞRk sst ðpÞ

(9.61)

Next, substituting (9.61) into (9.60) leads to 1 Qðb a ; pÞ ¼ xH k Rk xk

2 1 jxH k Rk sst ðpÞj 1 sH st ðpÞRk sst ðpÞ

(9.62)

Differentiating (9.62) with respect to p and setting the result to zero yields the maximum likelihood estimate (MLE) of p, given as b p . Consequently, the estimator is [33] 2 3 b ! f 2 1 H 6 7 js ðpÞR x j k k st arg max (9.63) pb ¼ 6 q7 1 4b 5¼ sH st ðpÞRk sst ðpÞ p¼½ f q fd T b fd Observe that (9.63) is very sensible, taking the form of the normalized magnitude of the STAP output. Implementing the estimator requires a very fine grid search to find the peak of the likelihood function (via the MLE approach). The grid search – which essentially amounts to stepping the space–time steering vector over potential target angles and Dopplers – is computationally burdensome and can be suboptimum. Specifically, if the step size is too large, the estimator can exhibit bias and increased variance. For this reason, numerical approximations to the MLE, which exhibit good practical performance, are commonly used [e.g., polynomial fits to the cost surface resulting from (9.63)]. Figure 9-23 shows an example of the minimum variance distortionless response spectrum from [34] for a generic X-band radar simulation. This figure shows the clutter ridge along the diagonal and an interference strobe in the vertical direction at a fixed angle. The angle is measured in the antenna coordinate system (ACS); the ACS definition is not important to the present discussion. Figure 9-24 shows the standard

9.8

Angle and Doppler Estimation

MVDR SPECTRA (dB) 30 4,000 25

Doppler Frequency (Hz)

3,000 2,000

20 1,000 0

15

423 FIGURE 9-23 ¢ Example AngleDoppler Interference Spectrum for Airborne STAP, Where the Vertical Line Corresponds to RFI [after [34], † 2004 IEEE].

–1,000 10

–2,000 –3,000

5

–4,000 –5,000 50

55

60 65 Angle (Degrees, Reverse ACS)

70

STD MLE Angle

6

4,000 5

Doppler Frequency (Hz)

3,000 2,000

4

1,000 0

3

–1,000 2

–2,000 –3,000

1

–4,000 –5,000 56

58 60 62 64 Angle (Degrees, Reverse ACS)

66

deviation of the MLE of (9.63) through Monte Carlo analysis for each unambiguous angle and Doppler observable by the radar. As seen from Figure 9-24, the MLE gives very good estimation performance away from the clutter ridge and interference strobe; this makes sense, as we expect better performance in regions where the SINR is higher. In the PDI case, it can be shown that the MLE amounts to summing the corresponding cost surfaces of the form of (9.63) for each CPI and then finding the peak. A similar derivation for the case of uncorrelated target voltages due to multistatic geometry is given in [35] and applies to the PDI case given in Figure 9-22, where frequency is instead used to decorrelate target voltages from CPI to CPI.

FIGURE 9-24 ¢ Standard Deviation of Monte Carlo MLE Direction of Arrival Estimates, in Degrees [after [34], † 2004 IEEE].

424

CHAPTER

9.9

9

Surface Moving Target Indication

OTHER CONSIDERATIONS

9.9.1 Heterogeneous Clutter The severity of the surface clutter environment influences SMTI performance. As previously discussed, adaptive radar techniques provide significant advantage over their nonadaptive counterparts. STAP estimates clutter properties from a training interval. Similarly, CFAR algorithms estimate the residual clutter and interference power levels prior to applying a threshold multiplier. A common assumption of most adaptive algorithms is independence and statistical similarity among the elements of the training set and the null-hypothesis condition of the application region. STAP uses training data to estimate the unknown, null-hypothesis covariance matrix, commonly using a sample covariance formulation similar to the calculation in (9.45) with the appropriately formatted data vectors. This approach is developed extensively in [16]. An underlying assumption of the approach is that the training data are multivariate Gaussian, as well as independent and identically distributed. When the iid assumption holds, the expected value of the covariance matrix tends to the actual covariance matrix. The real world, however, is generally not iid. The non-iid behavior is a result of the heterogeneous nature of the clutter environment. Sources of clutter heterogeneity include spatially varying clutter reflectivity, spatially varying clutter spectral spread, clutter discretes, moving targets corrupting the training data, extensive shadowing, and clutter-region boundaries. The impact of heterogeneous clutter on adaptive radar performance is discussed at length in [14, 19, 36, 37]. Heterogeneous clutter leads to covariance matrix estimation error, and consequently, filter mismatch relative to the null-hypothesis condition of the cell under test (note, the covariance error is not due to finite amounts of training data but to the fact the training data are non-iid). Solutions to mitigate the impact of heterogeneous clutter on adaptive radar are available from a number of sources. The driver for the array of CFAR algorithms has been heterogeneous clutter. In the case of STAP, the solution space includes training techniques, constraints, and knowledge-aided implementations [14, 38].

9.9.2 SAR-GMTI Spatial aperture is a major cost driver in radar-system design. In addition, the radarbearing platform constrains the aperture dimensions. Extending the temporal dwell is a potential approach to enhance detection of slow-moving targets at extended range. Interest also exists in moving target detection as a coproduct to a typical SAR mode. As the dwell increases, however, uncompensated target motion effects lead to significant matched filter losses, as Figure 9-15 indicates. For this reason, an effective approach to detection blends multichannel processing to suppress stationary clutter with an estimate of the nonlinear target phase history to accomplish the linearly transformed matched filtering [27, 28]. Assuming the target motion is adequately given by the combination of radial and tangential velocities, with origin specified at the antenna, the time-varying range to the moving target is expressible as rðtÞ ¼

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 ðv2p 2vp vT =t Þt2 þ 2ro vT =r t ro2 þ ðv2p 2vp vT=t Þt2 þ 2ro vT =r t ro þ (9.64) 2 ro

9.9

Other Considerations

where ro is the starting range to the target from aperture center, vp is the platformvelocity vector in the along-track direction, vT/t is the target’s tangential velocity, and vT/r is the target radial velocity. The target phase history is then u(t) ¼ 4pr(t)/l, where the quadratic term defines a chirp slope due to the tangential target motion. This quadratic term results in target energy smearing across Doppler bins. The linear term in (9.64) defines the target range rate; this differs from typical SAR processing, where Doppler determines cross-range position. Rather, for moving targets, the system requires multiple receive channels to measure physical direction of arrival. Acceleration leads to higherorder phase terms, up to fourth order, that greatly complicate searching for a suitable match to the range of plausible target hypotheses.

9.9.3 Bistatic and Multistatic Configurations The bistatic configuration consists of a transmitter separately located a significant distance from the receiver. In the case in which both transmitter and receiver are airborne, substantial separation between the transmitter and receiver leads to distinct bistatic phenomenology. In SMTI, the predominant challenge centers on the spatially varying bistatic clutter properties induced by the sensor geometry. Specifically, the angle-Doppler properties of bistatic clutter can vary from range bin to range bin; the angle of arrival is a function of the receive platform orientation, but both transmitter and receiver contribute to the Doppler frequency. Details of bistatic clutter nonstationarity are given in [39–43]. Clutter nonstationarity violates the iid assumption fundamental to STAP implementation. The covariance estimate represents the average properties of the data comprising the training interval. The corresponding covariance error leads to filter mismatch and, in some cases, greatly degraded detection performance. Fortunately, a number of effective bistatic STAP methods are available, as [40] summarizes. The bistatic STAP methods generally fall into three categories: localized training, data warping, or timevarying weights. In the localized training method, the STAP implementation selects training data in the vicinity of the cell under test in an attempt to minimize clutter-range variation. The data-warping approach modulates the data in a pre-processing step to create a degree of similarity to a designated reference; the modulation can be in time [43] or space–time [39]. In the time-varying weight approach, the weights are expanded using a truncated Taylor series, typically to first order, to track range variation in the clutter properties [42].

9.9.4 Dismount Detection Detecting personnel on Earth’s surface is the objective of dismount radar [44]. Dismount detection at suitable ranges is a challenge. The dismount RCS is over an order of magnitude smaller than that of a typical vehicle target. In addition, the dismount energy smears across Doppler filters, as the spectrogram in Figure 9-25 indicates. The sinusoidal variation in Figure 9-25 corresponds to the torso motion, whereas the other frequency components are due to movement of the extremities. The spectrogram is generated from a series of overlapping, short time Fourier transforms; for the example shown, the single pulse SNR is strong. When clutter is present and the radar is searching for targets at greater range, the detection processor must better match to the dismount target phase history to boost the SINR to levels acceptable for a useful detection rate.

425

CHAPTER

Surface Moving Target Indication

Velocity (m/s)

FIGURE 9-25 ¢ Simulated Spectrogram of a Dismount Taking Two Walking Steps at an Average Speed of 1.85 m/s [after [44], † 2008 IEEE].

9

6

20

5

15

4

10

3

5

2

1

–5

–10

–1 –1

9.10

–0.5

0 Time (s)

0.5

1

SNR (dB)

426

–15

SUMMARY

SMTI radar design and implementation is an important component of modern radar. This chapter discussed the fundamentals of SMTI. SMTI is a radar mode with a history that started in the early 1990s with Joint STARS. Since that time, interest in SMTI has grown considerably, with efforts to deploy SMTI on manned aircraft, UASs, and satellites. WAS-SMTI involves searching considerable areas of Earth’s surface by rapidly scanning the transmit beam and receive subarrays, processing considerable amounts of data in short time periods, and displaying the evolving target dots to the operator. In contrast, selective search and persistent search SMTI focus the radar system on smaller surface areas, leveraging the additional radar timeline to enhance detection of specific targets of interest. The impact of clutter returns on detection is a limiting factor in SMTI radar. Clutter mitigation requires careful system design and signal processing. In particular, the along-track length of the radar antenna, relative to the platform velocity, is a critical consideration. Longer arrays yield lower minimum detectable velocities for a fixed platform velocity. Sophisticated processing methods are further necessary to mitigate clutter effects. STAP provides the capability to see targets to within a fraction of the mainbeam clutter spread, with an improvement generally exceeding a factor of 2 to 5, depending on the circumstances. At its essence, SMTI radar attempts to discriminate the angle-Doppler response of a potential target from the background clutter. This chapter describes antenna, waveform, and signal-processing issues in considerable detail. In addition, the fundamental clutter, RFI, receiver noise, and target models are described and then used in benchmark analyses to compare the performance of varying methods. The latter parts of this chapter describe the end-to-end detection processing chain and a standard approach to bearing and Doppler estimation. Discussion of some germane SMTI issues culminates the chapter: an overview of the impact of heterogeneous clutter on detection performance, long coherent dwell processing considerations (SARGMTI), bistatic and multistatic configurations, and dismount detection.

9.12

9.11

References

FURTHER READING

SMTI is an application of modern radar technology. As this chapter describes, key SMTI elements include a detailed understanding of clutter and target phenomenology, multichannel antenna design, and advanced signal processing. These topics are covered in multiple sources. The interested reader should find this chapter’s references to be very useful for further investigation. We highlight several of these references as good starting points. We recommend that the interested reader spend a few moments reviewing [1] for historical reasons. Also, a perusal of the NATO STANAG 4607 format documents, found at the NATO website, is worthwhile. STAP is an important SMTI enabling technology. The seminal paper by Brennan and Reed, given as [15], should be read by anyone interested in STAP, along with the companion paper by Reed, Mallett, and Brennan in [16] describing the most commonly used approach to estimate the unknown covariance matrix. The report by Ward in [20] is a clear exposition on the basics of STAP, and the STAP overview by Melvin in [25] summarizes key STAP concepts and research trends. Both are worth spending some time reviewing and provide good foundational discussion on STAP technology. The book on STAP by Guerci in [17] should also be considered further reading; this book provides many important STAP insights. It is probably best to read these suggested texts chronologically as [25], [20], and then [17]. An in-depth understanding of STAP is critical to a detailed appreciation for SMTI radar.

9.12

REFERENCES

[1] J.N. Entzminger, C.A. Fowler, and W.J. Kenneally, ‘‘JointSTARS and GMTI: Past, present and future,’’ IEEE Trans. AES, Vol. 35, No. 2, April 1999, pp. 748–761. [2] USAF official website, http://www.af.mil/information/factsheets/factsheet.asp?id¼13225. [3] DARPA website, http://www.darpa.mil/ipto/programs/vader/vader.asp. [4] ‘‘Lynx SAR/GMTI radar: Increased capability for manned and unmanned missions,’’ General Atomics Aeronautical factsheet, General Atomics Aeronautical Systems, Inc. 13322 Evening Creek Dr., N., San Diego, California 92128 USA. [5] Space Radar information, http://www.losangeles.af.mil/library/factsheets/factsheet.asp? id¼5308. [6] D. Rabideau and S. Kogon, ‘‘A signal processing architecture for space-based GMTI radar,’’ in Proc. 1999 IEEE Radar Conference, Waltham, MA, pp. 96–101. [7] ‘‘RADARSAT-2: A new era in synthetic aperture radar,’’ RADARSAT-2 factsheet, Geospatial Services, 13800 Commerce Parkway, Richmond, BC Canada, V6V 2J3. See www.RADARSAT2.info. [8] C.H. Gierull and C. Linvingstone, ‘‘SAR-GMTI concept for RADARSAT-2,’’ Chapter 6, The Applications of Space-Time Processing, R. Klemm (Ed.), IEE Radar, Sonar, Navigation and Avionics 9, IEE Press, 2004. [9] C.M Rader, ‘‘A simple method for sampling in-phase and quadrature components,’’ IEEE Trans. AES, Vol. AES-20, No. 6, November 1984.

427

428

CHAPTER

9

Surface Moving Target Indication

[10] D.H. Johnson and D.E. Dudgeon, Array Signal Processing: Concepts and Techniques, Prentice-Hall, Englewood Cliffs, NJ, 1993. [11] J.R. Guerci, ‘‘Theory and application of covariance matrix tapers for robust adaptive beamforming,’’ IEEE Trans. Signal Processing, Vol. 47, No. 4, April 1999, pp. 977–985. [12] M.I. Skolnik, Introduction to Radar Systems, 2nd Ed., McGraw Hill, New York, NY, 1980. [13] J.B. Billingsley, Low-Angle Radar Land Clutter: Measurements and Empirical Models, William Andrew Publishing, Inc., Norwich, NY, 2002. [14] W.L. Melvin and G.A. Showman, ‘‘Knowledge-aided parametric covariance estimation,’’ IEEE Trans. AES, July 2006, pp. 1021–1042. [15] L.E. Brennan and I.S. Reed, ‘‘Theory of adaptive radar,’’ IEEE Trans. AES, Vol. 9, No. 2, March 1973, pp. 237–252. [16] I.S. Reed, J.D. Mallett, and L.E. Brennan, ‘‘Rapid convergence rate in adaptive arrays,’’ IEEE Trans. AES, Vol. 10, No. 6, November 1974, pp. 853–863. [17] J.R. Guerci, Space-Time Adaptive Processing for Radar, Artech House, Norwood, MA, 2003. [18] B.M. Keel, ‘‘Fundamentals of pulse compression waveforms,’’ Chapter 20, Principles of Modern Radar: Basic Principles, M.A. Richards et al. (Eds.), SciTech Publishing, Inc., North Carolina, 2010. [19] B.C. Armstrong, H.D. Griffiths, C.J. Baker, and R.G. White, ‘‘Performance of adaptive optimal Doppler processors in heterogeneous clutter,’’ IEE Proc.-Radar, Sonar, Navig., Vol. 142, No. 4, August 1995, pp. 179–190. [20] J. Ward, ‘‘Space-time adaptive processing for airborne radar,’’ Lincoln Laboratory Tech. Rept., ESC-TR-94-109, December 1994. [21] W.L. Melvin, ‘‘Clutter suppression using space-time adaptive processing,’’ Principles of Modern Radar: Advanced Techniques, ed. W. Melvin and J. Scheer, Sci-Tech Publishing, Edison, NJ, 2011. [22] R. Blum, W.L. Melvin, and M.C. Wicks, ‘‘An analysis of adaptive DPCA,’’ in Proc. 1996 IEEE National Radar Conference, Ann Arbor, Michigan, May 13–16, 1996, pp. 303–308. [23] R. Klemm, Space-Time Adaptive Processing: Principles and Applications, IEE Radar, Sonar, Navigation and Avionics 9, IEE Press, London, UK, 1998. [24] R. Klemm, Principles of Space-Time Adaptive Processing, 2nd Ed., IEE Radar, Sonar, Navigation and Avionics 12, IEE Press, UK, 2002. [25] W.L. Melvin, ‘‘STAP overview,’’ IEEE AES Systems Magazine – Special Tutorials Issue, Vol. 19, No. 1, January 2004, pp. 19–35. [26] R.C. DiPietro, ‘‘Extended factored space-time processing for airborne radar,’’ in Proc. 26th Asilomar Conf., Pacific Grove, CA, October 1992, pp. 425–430. [27] J.H.G. Ender, ‘‘Space-time processing for multichannel synthetic aperture radar,’’ IEE Electronics & Comm. Eng. J., February 1999, pp. 29-37. [28] G.A. Showman and W.L. Melvin, ‘‘Multi-resolution processing to enhance knowledgeaided STAP’’, in Proc. 2003 DARPA/AFRL KASSPER Workshop, April 14–16, 2003, Las Vegas, NV. [29] M. Kirscht, ‘‘Detection and imaging of arbitrarily moving targets with single-channel SAR,’’ IEE Proc.—Radar Sonar Navig., Vol. 150, No. 1, February 2003, pp. 7–11. [30] R.P. Perry, R.C. DiPietro, and R.L. Fante, ‘‘SAR imaging of moving targets,’’ IEEE Trans. AES, Vol. 35, No. 1, January 1999, pp. 188–199.

9.12

References

[31] J.R. Fienup, ‘‘Detecting moving targets in SAR imagery by focusing,’’ IEEE Trans. AES, Vol. 37, No. 3, July 2001, pp. 794–809. [32] F.C. Robey, D.R. Fuhrman, E.J. Kelly, and R. Nitzberg, ‘‘A CFAR adaptive matched filter detector,’’ IEEE Trans. AES, Vol. 28, No. 1, January 1992, pp. 208–216. [33] R.C. Davis, L.E. Brennan, and I.S. Reed, ‘‘Angle estimation with adaptive arrays in external noise fields,’’ IEEE Trans AES, Vol. AES-12, No. 2, March 1976, pp. 179–186. [34] G.A. Showman, W.L. Melvin, and D.J. Zywicki, ‘‘Application of the Crame´r-Rao lower bound for bearing estimation to STAP performance studies,’’ in Proc. 2004 IEEE Radar Conference, Philadelphia, PA, 26–29 April 2004, ISBN No. 0-7803-8235-8. [35] W.L. Melvin, R. Hancock, M. Rangaswamy, and J. Parker, ‘‘Adaptive distributed radar,’’ in Proc. 2009 Int’l Radar Conf., Bordeaux, France, October 2009. [36] W.L. Melvin, ‘‘Space-time adaptive radar performance in heterogeneous clutter,’’ IEEE Trans. AES, Vol. 36, No. 2, April 2000, pp. 621–633. [37] W.L. Melvin, ‘‘STAP in heterogeneous clutter environments,’’ in The Applications of Space-Time Processing, ed. R. Klemm, IEE Radar, Sonar, Navigation and Avionics 9, IEE Press, 2004. [38] W.L. Melvin and J.R. Guerci, ‘‘Knowledge-aided sensor signal processing: a new paradigm for radar and other sensors,’’ IEEE Tran. AES, July 2006, pp. 983–996. [39] W.L. Melvin and M.E. Davis, ‘‘Adaptive cancellation method for geometry-induced nonstationary bistatic clutter environments,’’ IEEE Trans. AES, April 2007, pp. 651–672. [40] W.L. Melvin, ‘‘Adaptive moving target indication,’’ in Advances in Bistatic Radar, Chapter 11, ed. N. Willis and H. Griffiths, Sci-Tech Publishing, Rayleigh, NC, 2007. [41] B. Himed, J.H. Michels, and Y. Zhang, ‘‘Bistatic STAP performance analysis in radar applications,’’ in Proc. 2001 IEEE Radar Conf., Atlanta, GA, May 2001, pp. 198–203. [42] S.M. Kogon and M.A. Zatman, ‘‘Bistatic STAP for airborne radar systems,’’ in Proc. IEEE SAM 2000, Lexington, MA, March 2000. [43] F. Pearson and G. Borsari, ‘‘Simulation and analysis of adaptive interference suppression for bistatic surveillance radars,’’ in Proc. 2001 ASAP Symp., Lexington, MA, 13 March 2001. [44] R.K. Hersey, W.L. Melvin, and E. Culpepper, ‘‘Dismount modeling and detection from small aperture moving radar platforms,’’ in Proc. 2008 IEEE Radar Conf., May 2008, Rome, Italy.

429

CHAPTER

Space-Based SAR for Remote Sensing

10

Lee Blanton, General Atomics Aeronautical Systems, Inc., San Diego, CA

Chapter Outline 10.1 10.2 10.3 10.4 10.5 10.6 10.7 10.8 10.9

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Historical Perspective. . . . . . . . . . . . . . . . . . . . . . . . . Orbits. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Design Considerations for the Spaceborne SAR. . . . . Special Modes and Capabilities . . . . . . . . . . . . . . . . . Design Example: Germany’s TerraSAR-X . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

10.1

....... ....... ....... ....... ....... ....... ....... ....... .......

........ ........ ........ ........ ........ ........ ........ ........ ........

.... .... .... .... .... .... .... .... ....

431 438 451 457 473 482 493 494 498

INTRODUCTION

Space, as with land, sea, and air, is now a well-established operating environment for radars. As in other environments, radars operating in space assume a wide variety of forms and span a wide range of applications. This chapter will focus on satellite-borne synthetic aperture radars (SARs) for remote sensing applications. This is an important and growing field with numerous applications in the areas of oceanography; land use; seismology; volcanology; disaster assessment; and the monitoring of sea ice, maritime economic zones, environmental conditions, etc.

10.1.1 Spaceborne SARs Placing a SAR into orbit and operating it for many years involve considerable expense. To justify this expense, a spaceborne SAR must provide benefits that at least equal its cost. Spaceborne SARs have the following advantages over other types of sensors: ●

Ability to make observations through clouds.

Ability to make observations in darkness.

Ability to sense phenomena that other types of sensors cannot sense.

Near-global coverage repeated at regular intervals.

Long-term temporal coverage. 431

432

CHAPTER TABLE 10-1 ● ● ● ● ● ●

10 ¢

Space-Based SAR for Remote Sensing

Characteristics and Challenges of Spaceborne SARs

Long range; high two-way spreading loss (path loss) High platform velocity (large Doppler shift) Large antennas, some requiring distributed RF amplification and control circuitry Highly constrained PRF, often range-ambiguous Large imaged area resulting in a large data volume Image formation processing is usually done on the ground; system performance may be datalink-limited

Because the microwave wavelength is much longer than visible and infrared (IR) wavelengths, a SAR can observe phenomena that are either unobservable or much less evident to an electro-optic/infrared (EO/IR) sensor. Such phenomena include ocean wave patterns, surface roughness, soil moisture content, etc. The ability of a radar to sense the Doppler shift of the return signals also enables measurement of ocean and river currents and other phenomena involving motion. As with other long-range radars, achieving an adequate signal-to-noise ratio (SNR) is a major challenge. For this reason, all remote sensing SARs flown to date have operated in low earth orbit (LEO). Even from LEO, the long range between the radar and the imaged area results in a high free-space spreading loss (aka path loss) requiring fairly large antennas and moderately high radiated powers (1–5 kW peak). A high platform velocity imparts a large Doppler frequency shift to the surface returns. The Doppler bandwidth of the surface returns can be limited by using an antenna with a narrow azimuth beamwidth (i.e., an antenna that is long in the horizontal or along-track dimension). Both the azimuth and elevation beamwidths affect the choice of pulse repetition frequency (PRF) since the PRF is constrained by factors involving range ambiguities as well as the requirement to keep the Doppler shift of the mainlobe returns unambiguous. Because of these factors the PRF in a spaceborne SAR is likely to be comparable to the Doppler bandwidth of the mainlobe returns, making Doppler (azimuth) ambiguities a serious design issue. Waveform requirements (e.g., the PRF) and antenna beam requirements are therefore closely coupled and require careful tradeoffs during the system design process. Table 10-1 lists some of the top-level characteristics and challenges of spaceborne SARs as compared with airborne radars.

10.1.2 Other Types of Spaceborne Remote Sensing Radars Although the emphasis in this chapter is on spaceborne SARs, two other types of spaceborne radars—altimeters and scatterometers—are important in the field of remote sensing. These nonimaging instruments often share the same satellite bus with SAR instruments, providing measurements that are both complementary to and time-correlated with the SAR images, so they will be discussed briefly here. Radar Altimeters. Satellite-borne radar altimeters have made important contributions to the fields of geodesy and oceanography. Since the satellite orbit is usually known with great precision, a satellite-borne altimeter can measure small variations in the height of the ocean surface resulting from various geophysical phenomena. In the absence of any other disturbances, the equipotential surface corresponding to the local

10.1

Introduction

mean sea level is known as the geoid. Over the entire Earth the height of the geoid deviates from its mean value by –104 m to þ64 m primarily as a result of density variations in Earth’s mantle [1]. Superimposed on this ‘‘permanent’’ geoid topography are variations due to oceanic-scale circulation (currents), mesoscale circulation (eddies), regions of high or low atmospheric pressure, bathymetry (depth), wind, and tides. These variations range in magnitude from a few centimeters to a few meters. The effects of tides and wind are more pronounced in shallow water such as over continental shelves. Sea surface height variations due to tides and wind-driven waves must be averaged out of radar altimetric (range) measurements in order for variations due to other phenomena to be quantified. On the other hand, since wind waves increase the surface roughness and reduce the backscatter coefficient in the nadir direction, the amplitude of the nadir return can be used to infer the surface wind speed [1, 2]. Several design approaches have been used in spaceborne radar altimeters. Pulselimited altimeters use the pulse width (usually the compressed pulse width, i.e., the range bin width) to limit the lateral extent of the nadir-centered clutter patch where the altimetric measurement is desired. Beam-limited altimeters rely on the beamwidth rather than the range bin width for this purpose. Whether a pulse-limited or beam-limited approach is more desirable depends on the hardware capabilities (waveform generator, A/D converter, etc.), the antenna size, and the operating frequency. Other approaches have also been proposed. A ‘‘synthetic aperture’’ altimeter uses Doppler processing to provide finer resolution in the along-track dimension, while resolution in the acrosstrack direction is still limited by either the beamwidth or the range bin width. In some applications, particularly over land, a single nadir-looking beam does not provide sufficient coverage. In these cases the beam can be scanned in the across-track direction to cover a wider swath on either side of the satellite’s ground track. Well-calibrated radar altimeter instruments can provide a measurement precision of a few centimeters [2]. A fairly high frequency (13 GHz) is often used in order to obtain a narrow beamwidth with a reasonably sized antenna. Scatterometers. Scatterometers are relatively coarse-resolution, nonimaging radar instruments that measure the backscatter coefficient of the surface. They often employ long antennas producing fan beams that provide narrow angular resolution in one dimension. In other dimensions, fine resolution can be obtained either in range through the compressed pulse width or in Doppler through Doppler filtering. Over oceanic areas, measurements of the backscatter coefficient in different, orthogonal directions from the radar can be used to infer the surface wind speed. Winddriven waves produce a maximum backscatter return when they are viewed along the wind direction (i.e., perpendicular to the wave fronts) and a minimum return when viewed perpendicular to the wind direction (parallel to the wave fronts) [1, 2]. At incidence angles well away from nadir, the overall strength of the returns is proportional to the wind speed, which is opposite to the relationship observed in nadir-pointing altimeters. Surface wind speed is an important meteorological parameter that can be retrieved by scatterometers from mid-ocean areas where there are no surface instruments. Scatterometers designed for this purpose are often called ‘‘wind scatterometers,’’ although the wind measurement is indirect. Scatterometers are also used to observe land areas. Since scatterometers have resolutions of tens of kilometers, the most consistent results are obtained from fairly homogeneous terrain types, such as tropical forests. Tropical forests exhibit volumetric

433

434

CHAPTER

10

Space-Based SAR for Remote Sensing

scattering, making the backscatter coefficient relatively independent of the incidence angle [2]. For surfaces in which soil is at least partially visible, the backscatter coefficient can be related to the soil moisture content, although it is also affected by many other factors. Observations of changes in the backscatter coefficient of a particular area over time can be correlated with changes in the soil moisture conditions from in situ measurements for calibration purposes. Changes in soil moisture content and sea surface salinity can also be detected by passive microwave radiometers. Scatterometers and radiometers can, therefore, provide complementary data and each instrument can aid in calibrating the other.

10.1.3 List of Symbols a ¼ semi-major axis of the satellite orbit A ¼ aperture area Bd ¼ Doppler bandwidth Bn ¼ noise bandwidth B0 ¼ Earth’s magnetic field strength 0.5 gauss near Earth’s surface Bx ¼ along-track phase center spacing c ¼ propagation velocity 3.0 108 m/s f ¼ frequency fd ¼ Doppler frequency shift fr ¼ pulse repetition frequency (PRF) F ¼ noise figure Gr ¼ receive antenna gain gs ¼ acceleration of gravity at Earth’s surface 9.81 m/s2 Gt ¼ transmit antenna gain H ¼ antenna aperture height (vertical dimension) hsat ¼ height of the satellite above the mean Earth radius i ¼ inclination of the orbital plane k ¼ Boltzmann’s constant ¼ 1.38 1023 J/K or W/(HzK) kaz ¼ azimuth beamwidth coefficient kel ¼ elevation beamwidth coefficient L ¼ antenna aperture length (horizontal or along-track dimension) LSA ¼ synthetic array (or synthetic aperture) length Lsys ¼ system loss (total) Me ¼ mass of Earth ¼ 5.972 1024 kg msat ¼ mass of the satellite P ¼ orbital period Pavg ¼ average transmitted power

10.1

Introduction

Pt ¼ peak transmitted power Re ¼ radius of Earth (mean) ¼ 6,371.462 km rsat ¼ radius of the satellite from Earth’s center (instantaneous) s ¼ antenna element spacing R, Rs, Rs(min), Rs(max) ¼ slant range Tr ¼ pulse repetition interval (PRI) TSA ¼ synthetic array (or synthetic aperture) time T0 ¼ standard reference temperature ¼ 290 K vr ¼ radial (‘‘line-of-sight’’) velocity vsat ¼ satellite velocity b, bnear, bfar ¼ angle at the target location on Earth’s surface between the directions toward the satellite and the center of Earth bL ¼ argument of latitude g, gmin, gmax ¼ angle at the center of Earth between the directions toward the satellite and the target location dCR ¼ cross-range (‘‘azimuth’’) resolution dR ¼ range resolution Dq ¼ change in target aspect angle during the synthetic array time Dq0 ¼ shift in the beam position (‘‘beam squint’’) with a change in frequency DRg ¼ difference between the maximum and minimum ground (surface) ranges in the mainlobe footprint DRs ¼ difference between the maximum and minimum slant ranges in the mainlobe footprint j ¼ angle of propagation path from nadir ha ¼ antenna aperture efficiency q ¼ true anomaly (angular position of the satellite in its orbit, measured from the periapsis) qb ¼ one-way half-power beamwidth qb(Az) ¼ one-way half-power azimuth beamwidth qb(El) ¼ one-way half-power elevation beamwidth qB ¼ angle between the propagation path and Earth’s magnetic field qg ¼ angle of the first grating lobe from broadside qi ¼ angle of incidence of the wavefront at Earth’s surface qL, qL(min), qL(nom), qL(max) ¼ look angle from nadir qs, qs(max) ¼ antenna beam scan angle l ¼ wavelength m ¼ gravitational parameter ¼ 3.986 1014 m3/s2 for Earth s ¼ radar cross section (RCS)

435

436

CHAPTER

10

Space-Based SAR for Remote Sensing

sc ¼ clutter RCS in a resolution cell s0 ¼ surface backscatter coefficient (area-normalized RCS) s0N ¼ noise-equivalent sigma zero (NESZ), also known as noise-equivalent sigma ‘‘naught’’ (NEs0) or additive noise coefficient (ANC) t ¼ pulse width (uncompressed) y ¼ grazing angle w ¼ argument of the perigee/periapsis we ¼ Earth’s angular rotation rate wsat ¼ satellite angular rate W ¼ right ascension of the ascending node WP ¼ polarization rotation angle

10.1.4 List of Abbreviations and Acronyms ADC – Analog-to-Digital Converter AESA – active electronically scanned array AFB – Air Force Base aka – also known as ALOS – advanced land observation satellite ANC – additive noise coefficient, also known as noise-equivalent sigma zero (NESZ) ASAR – advanced synthetic aperture radar ATI – along-track interferometry BAQ – block adaptive quantizer BFPQ – block floating point quantization C-band – 4–8 GHz (typically ~5.3 GHz for spaceborne SARs) COSMO – Constellation of Small Satellites for Mediterranean Basin Observation DARPA – Defense Advanced Research Projects Agency DC – direct current; also zero Doppler frequency DEM – digital elevation measurement DPCA – displaced phase center antenna DTAR – distributed target ambiguity ratio EO – electro-optic ERIM – Environmental Research Institute of Michigan ESA – European Space Agency FBAQ – flexible block adaptive quantizer FM – frequency modulation GMTI – ground moving target indicator

10.1

Introduction

HH – transmit using horizontal polarization, receive using horizontal polarization HV – transmit using horizontal polarization, receive using vertical polarization I – inphase IR – infrared ITU – International Telecommunication Union JAROS – Japan Resources Observation System Organization JERS – Japanese Earth Resources Satellite JPL –Jet Propulsion Laboratory L-band – 1–2 GHz (typically ~1.3 GHz for spaceborne SARs) LEO – low earth orbit LFM – linear frequency modulation mrad – milliradians MTI – moving target indicator NASA – National Aeronautics and Space Administration NASDA – National Space Development Agency of Japan NESZ – noise-equivalent sigma zero PALSAR – phased array type L-band SAR PRF – pulse repetition frequency PRI – pulse repetition interval Q – quadrature RCS – radar cross section SAR – synthetic aperture radar S-band – 2–4 GHz (typically ~3.1 GHz for spaceborne SARs) SCR – signal to clutter ratio SDNR – signal to distortion noise ratio SIR-A – Shuttle Imaging Radar - A SIR-B – Shuttle Imaging Radar - B SIR-C – Spaceborne Imaging Radar - C SNR – signal to noise ratio SRTM – Shuttle Radar Topography Mission STAP – space–time adaptive processing STS – space transportation system TanDEM-X – TerraSAR-X add-on for digital elevation measurements TEC – total electron content T/R – transmit/receive TRM – transmit/receive module TRMM – Tropical Rainfall Measuring Mission

437

438

CHAPTER

10

Space-Based SAR for Remote Sensing

U.S. – United States VV – transmit using vertical polarization, receive using vertical polarization VH – transmit using vertical polarization, receive using horizontal polarization X-band – 8–12 GHz (typically ~9.5 GHz for spaceborne SARs) X-SAR – X-Band synthetic aperture radar (shuttleborne)

10.2

HISTORICAL PERSPECTIVE

10.2.1 Overview The use of SARs for civil remote sensing began with the NASA/Jet Propulsion Laboratory (JPL) Seasat mission in 1978. Seasat carried three pioneering radar instruments for studying Earth’s oceans: an L-band SAR, a precision radar altimeter, and a microwave scatterometer. Since Seasat many other SAR satellites have successfully flown. The later satellites have operated over a wider range of frequencies (L-band through X-band), have provided multipolarization measurements, and have implemented other capabilities, such as electronic scanning, that continue to evolve over time. The U.S. Seasat mission of 1978 was followed by five short-duration shuttle imaging radar (SIR) missions of increasing sophistication between 1981 and 2000: SIR-A, SIR-B, SIR-C/XSAR (two flights), and the Shuttle Radar Topography Mission (SRTM). The United States has not launched a free-flying civil SAR satellite since Seasat. In 1987 and 1991 the USSR orbited two large SAR spacecraft (Cosmos 1870 and Almaz-1). These satellites were both unmanned versions of the Salyut spacecraft carrying S-band (3.1 GHz) synthetic aperture radars. They functioned for 1.5 to 2 years, after which they were intentionally deorbited. Russia has not orbited any civil SAR satellites since that time but several new programs have reportedly been under consideration. Between 1991 and 2006, spaceborne SARs were developed by the European Space Agency (ESA) and the space agencies of Japan and Canada. The European Remote Sensing satellite ERS-1 carried a C-band SAR and operated from 1991 through 2000. Follow-on missions ERS-2 and ENVISAT also carried SARs operating in the C-band. The Japanese Earth Resources Satellite JERS-1 operated in L-band from 1992–1998. Canada has operated the highly successful RADARSAT-1 carrying a C-band SAR since its launch from California in 1995 (still operating in early 2013). A more sophisticated follow-on satellite, RADARSAT-2, was launched for Canada by Russia in 2007. Beginning in 2006 a new generation of advanced SAR satellites appeared, including RADARSAT-2, with the majority of the others being developed and operated by Germany and Italy. Germany placed its five-satellite SAR-Lupe constellation into orbit between 2006 and 2008 using Russian launch services. These satellites operate in X-band for the purpose of security-related reconnaissance. Germany also operates two civil remote sensing SAR satellites called TerraSAR-X, launched in 2007, and its neartwin TanDEM-X, launched in 2010. These satellites orbit in close formation, enabling the pair to produce highly accurate elevation maps through the use of interferometry. Italy’s four-satellite SAR constellation, COSMO/Skymed, was placed into orbit between 2007 and 2010 using U.S. launch vehicles from Vandenberg AFB, California. These X-band satellites form a dual-use system providing imagery for both security-related reconnaissance and civil remote sensing purposes.

10.2

Historical Perspective

Since 2008 India and Israel (using an Indian launch vehicle) have orbited their own SAR satellites—Israel’s TecSAR and India’s RISAT-1 and RISAT-2. ESA, Canada, and Japan continue to evolve their SAR remote sensing capabilities with future launches planned. In addition, China and Japan have spaceborne SAR programs supporting their national security. Several other countries, notably South Korea and Argentina, have SAR remote sensing satellites under development. A timeline of SAR remote sensing missions since 1978 is presented in Figure 10-1. Table 10-2 presents technical characteristics of some of the more historically significant SAR missions, although space limitations prevented the inclusion of many other interesting programs. Brief descriptions of a few of the more significant and widely publicized missions are presented in the subsections that follow.

10.2.2 U.S. Spaceborne SARs Seasat (1978). Seasat was the first civil spaceborne SAR. It was developed by the Jet Propulsion Laboratory under funding from NASA. Seasat was launched in June 1978 into a circular orbit with an altitude of 800 km and an orbital inclination of 108 , completing 14 orbits per day. Seasat carried three active radar instruments: an L-band synthetic aperture radar, a Ku-band radar altimeter, and a Ku-band scatterometer. It also carried two passive instruments: a visible and infrared radiometer and a scanning multichannel microwave radiometer. The Seasat SAR used horizontal polarization on transmit and horizontal polarization on receive (a combination designated HH) to monitor the global surface wave field and polar sea ice conditions. A low off-nadir look angle of only 20 provided adequate SNR and accentuated returns from ocean surface features. Seasat operated for 105 days until October 1978 during which approximately 42 hours of SAR data were collected, an average of about 24 minutes per day [3, 4]. There was no onboard recorder, so imagery and other data could only be collected when the satellite was within range of a ground station. An artist’s depiction of the Seasat satellite is shown in Figure 10-2. Shuttle Imaging Radar SIR-A (1981). The Shuttle Imaging Radar A (SIR-A) was flown aboard the Space Shuttle Columbia in November 1981 on the second shuttle flight. SIR-A, as well as its successors SIR-B and SIR-C/X-SAR, operated from the shuttle payload bay and did not fly freely. SIR-A was largely assembled from spare parts from the Seasat program [5]. As compared with Seasat, SIR-A operated at a lower altitude of 259 km and a lower orbit inclination of 38 due to shuttle flight constraints. SIR-A had a narrower system bandwidth (6 MHz) than that of Seasat and a coarser resolution of 40 m in azimuth and range. The SIR-A swath width of 50 km was narrower, and the incidence angle of 503 was considerably higher (i.e., the grazing angle at the surface was shallower) than that of Seasat [6]. The higher incidence angle was more suitable for imaging terrain with significant relief as opposed to the ocean surface. Radar data was recorded optically on film for later processing on the ground. Shuttle Imaging Radar SIR-B (1984). SIR-B was an updated radar using newly designed hardware that was flown aboard the Space Shuttle Challenger on flight 41-G (shuttle flight 13) in October 1984. The orbit was nominally circular during radar data collection, although several different altitudes were used. The average altitude for the first 20 orbits was 360 km, for the next 29 orbits it was 257 km, and for the remainder of the mission it was 224 km. These altitudes were fairly comparable with those of the

439

¢

Jan-93

Jan-98

Jan-03

Timeline of SAR Satellite Missions.

Jan-88

Jan-08

Jan-13

10

FIGURE 10-1

Jan-83

CHAPTER

Jan-78

USA Seasat SIR-A SIR-B SIR-C/X-SAR SIR-C/X-SAR SRTM Russia COSMOS-1870 ALMAZ Canada RADARSAT-1 RADARSAT-2 RADARSAT Constellation-1 RADARSAT Constellation-2 RADARSAT Constellation-3 ESA ERS-1 ERS-2 Envisat Sentinel-1A Sentinel-1B Japan JERS ALOS ALOS-2 Germany SAR Lupe-1 SAR Lupe-2 SAR Lupe-3 SAR Lupe-4 SAR Lupe-5 TerraSAR-X TanDEM-X TerraSAR-X2 Italy COSMO-Skymed-1 COSMO-Skymed-2 COSMO-Skymed-3 COSMO-Skymed-4 Israel TECSAR India RISAT-2 RISAT-1 Korea KOMPSAT-5 Argentina SAOCOM 1A SAOCOM 1B

440

Space-Based SAR for Remote Sensing

38 1.275 GHz (L-band)

108 1.275 GHz (L-band)

Look angle from nadir Incidence angle at the surface

Swath width (in range)

Antenna type Antenna dimensions Beamwidths (Az El)

Orbit inclination Radar frequency

259 km

800 km

Orbit altitude

1.27 6.24 (uniform wtg.) 50 km

47 503

1.26 6.24 (uniform wtg.)

20

23 3 across the swath

100 km

Passive array 9.4 m 2.16 m

Passive array 10.74 2.16 m

U.S. Nov. 1981 (2 days)

U.S. 1978 (105 days)

Nationality Mission dates (and duration)

Shuttle Imaging Radar A (SIR-A)

225 km

U.S. Apr. and Oct. 1994

Shuttle Imaging Radar C (SIR-C)/ X-SAR

15–90 km (L & C) 15–40 km (X)

El: 5.2–18.4

14 –56 17 –63 (at 257 km alt.) 15 –65 20 –65

1.26 6.24 (uniform wtg.) 20–40 km

57 1.277 GHz (L) 5.172 GHz (C) 9.677 GHz (X) Passive array AESA 10.7 m 2.16 m 12 m 4 m (overall)

360, 257, and 224 km 57 1.275 GHz (L-band)

U.S. Oct. 1984 (8.3 days)

Shuttle Imaging Radar B (SIR-B)

SAR Satellite and Radar Characteristics

Seasat

¢

Parameter

TABLE 10-2

20.1 –25.9

~20

100 km

0.29 5.4

Passive array 10 m 1 m

98.5 5.3 GHz (C-band)

Europe ERS-1: 1991–2000 ERS-2: 1995–2011 782–785 km

ERS-1 and ERS-2

15 –45

13 –39

100 km

0.29 2.2 (uniform wtg.)

AESA 10 1.3 m

98.55 5.331 GHz (C-band)

800 km

Europe 2002–2012

ENVISAT (ASAR instrument)

98.16 1.270 GHz (L-band)

690 km

Japan 2006–2011

ALOS PALSAR

AESA AESA 11.9 m 2.2 m 8.8 m 3.1 m 1.36 3.87 1 5.5 (uniform wtg.) 75 km 40–70 km; 250–350 km using ScanSAR 35 9.9 –50.8 (typical) 32 –38 8 –60 (24 and 39 typ.)

97.7 1.275 GHz (L-band)

568 km

Japan 1992–1998

JERS-1

20 –60

0.19 1.92 (uniform wtg.) 45–170 km; 300–500 km using ScanSAR 20 –59

Passive array 15 m 1.5 m

98.6 5.3 GHz (C-band)

793– 821 km

Canada Nov. 1995– 2013

Radarsat-1

(Continues )

10 –60

18–170 km; 300–500 km using ScanSAR 9 –50

Az: 0.21– 0.63

AESA 15 m 1.37 m

98.6 5.405 GHz (C-band)

793–821 km

Canada Dec. 2007– 2013

Radarsat-2

20–30 m

4 30.4 ms

25 m (Theoretical) 40 m

6 30.4 ms 1,464–1,824

4 33.4 ms

1,463–1,640

Transmitted peak power Transmit duty factor Data recorder bit rate

58–16 m

25 m (Theoretical) 40 m

Range resolution (cross-track) Azimuth resolution (along-track) Number of looks Transmitted pulse length PRF (Hz)

4.5–5.5% (est.)

4.9–5.5% (est.)

110 Mbits/s Optical film (on the ground) recorder

1.0 kW

1.0 kW

6 MHz

5.6–6.9%

4 8.5–33.8 ms (L, C) 40 ms (X) 1,395–1,736 (L, C) 1,240–1,860 (X) 3.3 kW (X)

30 m (L, C, X)

L & C band: HH, HV, VV, VH X band: VV 10 and 20 MHz (L, C) 9 and 19.6 MHz (X) 30 m (L, C) 10 and 20 m (X)

Shuttle Imaging Radar C (SIR-C)/ X-SAR

45–90 Mbits/s Optical film recorder (30.4 Mbits/s on ground)

1.12 kW

12 MHz

HH

19 MHz

HH

Waveform bandwidth

Shuttle Imaging Radar B (SIR-B)

HH

Shuttle Imaging Radar A (SIR-A)

Polarization

(Continued )

Seasat

¢

Parameter

Table 10-2

105 Mbits/s

6.2%

4.8 kW

1,640–1,720

4 37.12 ms

30 m

26.3 m

100 Mbits/s

3.5–4.5%

1.395 kW

30 m

30 m

HH, HV,VV, VH

VV

15.55 MHz

ENVISAT (ASAR instrument)

ERS-1 and ERS-2

60 Mbits/s

5.3–5.6%

1.3 kW

1,506–1,606

3 35 ms

18 m

18 m

15 MHz

HH

JERS-1

Radarsat-1

12– 100 MHz

3 m (min.)

11.6, 17.3, 30 MHz 5 m (min.)

HH, HV,VV, VH

Radarsat-2

120 and 240 Mbits/s

2.4–6.8%

2 kW

1,500–2,500

74–105

5.3–5.8%

5 kW

1,270–1,390

400 Mbps

10%

2.4–3.7 kW

1,300–3,800

8–100 m 1–100 m (30 m typical) (25–40 m typical) 2 1–16 1–4 16 and 27 ms 42 ms 21 and 42 ms

7–89 m (10–30 m typ.) 10 m

14 and 28 MHz

HH, HV,VV, HH VH

ALOS PALSAR

10.2

Historical Perspective

443 FIGURE 10-2 NASA Seasat Satellite [Jet Propulsion Laboratory].

SIR-A mission. The 57 orbit inclination was greater than that used for SIR-A, providing greater coverage of Earth’s surface. The system bandwidth of 12 MHz was twice that of SIR-A. A mechanically tiltable antenna provided a wide range of off-nadir look angles from 15 to 65 . This permitted observations of both oceans and land, as well as repeated observations of the same target area at multiple incidence angles [7]. SIR-C/X-SAR (1994). The Spaceborne Imaging Radar C/X-Band Synthetic Aperture Radar (SIR-C/X-SAR) was flown twice in 1994 on the Space Shuttle Endeavour, during April on flight STS-59 and during late September and early October on flight STS-68. (Note: The STS numbering did not follow a strict chronological sequence during this period.) The SIR-C portion of the system evolved from JPL’s L-band SIR-A and SIR-B radars, while the X-SAR X-band radar subsystem was provided by Germany [8]. Figure 10-3 shows the general concept of the SIR-C/X-SAR installation in the shuttle payload bay along with the antenna layout and dimensions.

10.2.3 European Space Agency (ESA) Spaceborne SARs ERS-1 and ERS-2. The European Remote Sensing satellites, ERS-1 and ERS-2, were launched in July 1991 and April 1995, respectively. Their near-polar sun-synchronous orbits had a mean altitude of approximately 780 km, an inclination angle of 98.5 , and an orbital period of 100 minutes. (A sun-synchronous orbit is an orbit in which the orbital plane maintains a relatively constant orientation with respect to the sun. With respect to the fixed stars, the plane of a sun-synchronous orbit precesses exactly 360 in one year, or slightly less than one degree per day. Sun-synchronous orbits will be discussed in greater detail in Section 10.3.) ERS-2 was initially placed into the same orbit as ERS-1 in a so-called tandem mission providing opportunities for SAR interferometry [9]. ERS-2 coverage of any particular point on the ground was phased one day behind

¢

444

CHAPTER

Space-Based SAR for Remote Sensing

10

FIGURE 10-3 ¢ SIR-C/X-SAR Antenna Installation, Dimensions, and Layout [eoportal.org].

Note: The Payload was Located Underneath the SAR Antenna The Shuttle Flew Upside Down for Observations All Dimensions are Given in Meters C–Band Microstrip Antenna Panel L–Band Microstrip Antenna Panels 1.3

0.75

4.1

2.95

0.4 4.0 12.0 m X-band Slotted Waveguide Antenna Panel Front View of Antenna Panels

that of ERS-1 [10]. The ground track of both ERS satellites repeated every 35 days. Both ERS-1 and ERS-2 SARs operated in C-band. Major instruments onboard ERS-2 included a synthetic aperture radar; a radar altimeter; a wind scatterometer; and several passive instruments covering microwave, IR, visible, and UV wavelengths. ERS-1 failed in March 2000 after operating for over nine years, which was over three times its design lifetime. ERS-2 operated for 16 years until September 2011 when it was taken out of service. Figure 10-4 shows the ERS-2 satellite; ERS-1 was very similar in appearance. Envisat ASAR. ESA’s Envisat satellite was launched in March 2002 and operated for ten years until April 2012. As shown in Figure 10-5, Envisat instruments included the advanced synthetic aperture radar (ASAR) with an active phased array antenna [11, 12]. At the 800-km orbit altitude, Envisat circled Earth every 100 minutes. For most instruments, global coverage was provided every 3 days, with exact repeat coverage occurring every 35 days, which was the same as for ERS-2. This repeat coverage

10.2

Historical Perspective

445

FIGURE 10-4 ¢ ERS-2 Satellite. ERS-1 was Similar in Appearance. Major Components from Left to Right are the Solar Array, Spacecraft Body, SAR Antenna, and a Pair of Scatterometer Antennas. [European Space Agency].

AATSR MIPAS SCIAMACHY

MERIS

MWR Ka-Band Antenna

GOMOS

DORIS RA-2 Antenna LRR

X-Band Antenna

ASAR Antenna

Service Module Solar Array (not Shown)

provided images for multipass SAR interferometry, enabling interferometric image pairs to be taken at the same incidence angles. Envisat operated for double its original fiveyear design life. Like ERS-2, the ASAR operated in C-band. While ERS-1 and ERS-2 used vertical polarization, ASAR had the capability to transmit either horizontal or vertical polarization, and to receive either horizontally or vertically polarized returns, or both, providing full coverage of the polarization matrix (HH, HV, VH, and VV). In the imaging and dual-polarization modes, the ASAR peak radiated power was 1,395 W.

FIGURE 10-5 ¢ Envisat Satellite and ASAR Antenna [European Space Agency].

446 FIGURE 10-6 ¢ ENVISAT ASAR Antenna Layout.

CHAPTER

10

Space-Based SAR for Remote Sensing 10 m

2m “Panel”: 2 m × 1.3 m

1m 0.65 m

Row: “Tile”: 16 Rows per Tile 20 Tiles 1 T/R Module per Row per Antenna 24 Elements per Row

1.3 m

Total Tiles = 20 Total T/R Modules: 20 × 16 = 320 Total Radiating Elements: 7,680

In contrast to ERS-1 and ERS-2, the Envisat ASAR used an active phased array antenna. This antenna was 10 m long by 1.3 m high and was composed of five ‘‘panels’’ that were each 2.0 m long by 1.3 m high. Each panel contained four ‘‘tiles’’ (antenna panel subassemblies) for a total of 20 tiles in the full array. Each tile was 1 m wide by 0.65 m high and was composed of 16 rows of 24 dual-polarized radiating elements. Each row was connected to a transmit/receive (T/R) module providing power amplification on transmit, low-noise amplification on receive, and amplitude and phase control at the subarray (row-in-tile) level. The ASAR antenna thus had a total of 320 transmit/receive modules driving 7,680 radiating elements, providing an electronic beam-steering capability in elevation and a more limited electronic beam-steering capability in azimuth [11, 13]. The ASAR antenna layout is shown in Figure 10-6. ASAR was also equipped with a programmable digital waveform generator. Another improvement compared to ERS-1 and ERS-2 was an eight-bit analog-to-digital converter (ADC) with a flexible block adaptive quantizer (FBAQ). The FBAQ provided 8/4 (i.e., 8 bits in, 4 bits out) and 8/2 compression ratios that enabled the collection of input signals spanning a larger dynamic range while remaining within the data rate constraints of the downlink. Block adaptive quantization is discussed in greater detail in Section 10.4.6.

10.2.4 Japanese Spaceborne SARs JERS-1. The Japanese Earth Resources Satellite (JERS) was launched in February 1992 into a 568-km circular orbit with a 97.7-degree inclination. It functioned for six years until October 1998 [14]. JERS-1 operated in L-band at an RF center frequency of 1.275 GHz with a peak transmitted power of 1.3 kW. The SAR resolution was 18 m. A 75-km range swath could be covered with three-look imagery. The corporate-fed antenna dimensions were 11.9 m by 2.2 m, and it was oriented to provide an incidence angle of approximately 35 at the surface. Most spaceborne SARs are able to operate for only a small portion of each orbit due to limitations in the data recording capacity and/or power storage (battery) capacity. The JERS-1 SAR could operate for up to 20 minutes per orbit. A total data rate of 60 Mbps was provided (30 Mbps per channel for I and Q channels), with three-bit quantization [15]. JERS-1 was a National Space Development Agency of Japan (NASDA) project. An illustration of the JERS-1 satellite is shown in Figure 10-7. ALOS PALSAR. The Phased Array Type L-band Synthetic Aperture Radar (PALSAR) development is a joint project between the NASDA and the Japan Resources Observation System Organization (JAROS). PALSAR is a payload on the advanced land

10.2

Historical Perspective

447 FIGURE 10-7 ¢ JERS-1 Satellite [Jet Propulsion Laboratory].

Data Relay Satellite Communication Antenna Star Trackers PALSAR

Solar Array Paddle

AVNIR-2 PRISM Velocity Nadir

observing satellite (ALOS), which was launched in January 2006 and functioned until April 2011. ALOS was in a sun-synchronous orbit at an altitude of 690 km with an orbital period of 99 minutes. The ALOS spacecraft configuration is illustrated in Figure 10-8 [17]. PALSAR operated in L-band at 1.27 GHz with waveform bandwidths of 14 and 28 MHz. PALSAR provided 10-m resolution with dual-polarization capability [16]. Like the ENVISAT ASAR, PALSAR employed an active phased array antenna. The deployed antenna dimensions were 8.8 m wide by 3.1 m high. It was organized into four panels that were each 2.2 m wide by 3.1 m high containing a total of 80 T/R modules [16]. The T/R modules were at the subarray level, so each T/R module served multiple radiating elements. The PALSAR antenna was electronically steerable in elevation with an off-nadir look angle capability of 9.9 to 50.8 .

10.2.5 Canadian Spaceborne SARs RADARSAT-1. Canada’s first radar satellite, RADARSAT-1, was launched in November 1995 from Vandenberg, AFB, California, into an approximately 800-km-high orbit

FIGURE 10-8 ¢ Japanese ALOS with PALSAR [European Space Agency].

448

CHAPTER

10

Space-Based SAR for Remote Sensing

inclined 98.6 . Launch services were provided by NASA in exchange for data collected by the satellite. RADARSAT-1 is still operating in early 2013, far exceeding its fiveyear design lifetime. RADARSAT-1 was placed into a sun-synchronous ‘‘dawn–dusk’’ orbit in which the orbital plane remains approximately aligned with the terminator (the boundary between Earth’s sunlit and dark sides). This provides nearly continuous solar illumination for power generation except during brief eclipse periods around the June solstice that affect a portion of the orbit at far southern latitudes. The radar normally looks toward the right of the satellite track in order to provide coverage of the region around the North Pole. Solar illumination is from the left. The satellite can occasionally be rotated 180 about its yaw axis to operate in a left-looking configuration providing temporary coverage of the South Pole. In that configuration, the radar looks toward the left of the satellite track and the solar illumination is still from the left. The SAR operates in C-band at 5.3 GHz, a frequency that is well suited for monitoring and categorizing sea ice. The peak transmit power is 5 kW with an average transmit power of 300 W [18]. RADARSAT-1 uses a planar array antenna providing seven selectable shaped elevation beams as shown in Figure 10-9 [18]. These beam patterns have peak gains varying from approximately 39 dBi for beam 1 (the broadest elevation beam) to approximately 42 dBi for beam 7 (the narrowest). Elevation beam shaping and beam selection or scanning are becoming increasingly important in spaceborne SARs for expanding the processed range swath using the ScanSAR mode (discussed in Section 10.5.4). The maximum RF bandwidth used by RADARSAT-1 is 30 MHz, providing a fine ground range resolution of about 9 m [18]. RADARSAT-1 also implements bandwidths of 17.3 MHz and 11.6 MHz providing coarser resolutions but with multilook capability and better image quality. The pulse repetition frequency (PRF) can be varied from 1,270 Hz to 1,390 Hz in 2-Hz steps [18]. The PRF is selected to avoid overlap between the desired mainlobe returns and the nadir sidelobe return (i.e., the ‘‘altitude return’’), which may be in a different range ambiguity, as well as to avoid eclipsing of the mainlobe returns [19]. The pulse length is 42.0 ms [20] resulting in a 5.3 to 5.8 percent duty cycle.

44.0 BEAM

1

2

3

4

5

6

7

42.0 ANTENNA GAIN (DB)

FIGURE 10-9 ¢ RADARSAT-1 SAR Elevation Beam Patterns [18].

40.0 38.0 36.0 34.0 32.0 30.0 10

20

30 ELEVATION ANGLE (°)

40

50

10.2

Historical Perspective

449

RADARSAT-2. RADARSAT-2 was launched from Baikonur, Kazakhstan, in December 2007 and was placed into the same orbit as RADARSAT-1. RADARSAT-2 operates in C-band at a center frequency of 5.405 GHz and provides a resolution down to 3 m in its ‘‘ultra-fine’’ resolution mode. The RADARSAT-2 SAR antenna, like the ASAR and PALSAR antennas, is an active phased array with electronic scan capability in elevation. It is 15 m long by 1.37 m high and is composed of a total of 512 subarrays arranged on two ‘‘wings.’’ This configuration allows the array to be partitioned into left and right halves to support alongtrack interferometry and other moving target detection modes. Each subarray is a row of 20 radiating elements associated with one transmit/receive (T/R) module providing amplification and phase control. Thirty-two subarrays are stacked to form a ‘‘column’’ or subpanel. Sixteen of these ‘‘columns’’ span the length of the antenna [21]. The horizontal center-to-center spacing between columns is 0.94 m, or 16.9 wavelengths. The center-tocenter spacing between the subarrays (rows of elements) in the height dimension (aka ‘‘width’’ as opposed to ‘‘length’’) is 0.77 wavelength, allowing the beam to be electronically scanned in elevation over a range of approximately 39 without incurring problems from grating lobes [20]. Figure 10-10 shows the layout of the RADARSAT-2 antenna along with the nomenclature used on that program. Each subarray can transmit either horizontal or vertical polarization and can receive both polarizations. Aperture weighting can be applied using phase-only weighting on transmit and using amplitude and/or phase on receive. Any subset of columns can be enabled or disabled on transmit or receive. The radar receiver provides two independent receive channels that can either be used for dual-polarized (horizontal and vertical) reception or for separate reception on the leading and trailing antenna wings to support along-track interferometry (discussed in Section 10.5.6). The linear frequency modulation (LFM) modulator can generate waveform bandwidths from 12 MHz to 100 MHz. The ultra-fine resolution mode provides approximately 3 m by 3 m resolution over 20-km swaths at incidence angles from 30º to 40º. To provide the additional aperture time required to achieve a 3-m cross-range resolution an azimuth beamwidth of approximately 0.5º is required. This can be achieved in the strip map mode either by using only a portion of the aperture or by slightly decollimating the beam in azimuth. In order to retain sensitivity, the T/R modules can be operated at a higher peak transmit power using a shorter pulse width in this mode as compared with other modes.

FIGURE 10-10 ¢ RADARSAT-2 Antenna Layout.

2 Wings; 4 Panels; 16 “Columns” 15 m “Wing” (2 Panels) “Panel” (4 “Columns”)

0.9375 m 1.37 m

“Column” (Subpanel): “Subarray” (Row of Elements): 16 “Columns” per Antenna 32 Subarrays per “Column” 1 T/R Module per Subarray 20 Elements per Subarray

Total “Columns”: 16 Total T/R Modules: 16×32 = 512 Total Radiating Elements: 10,240

450 TABLE 10-3

CHAPTER ¢

10

Space-Based SAR for Remote Sensing

COSMO-SkyMed Modes and Parameters [22]

Mode

Enhanced Spotlight

Himage Strip Map

PingPong Strip Map

ScanSAR Wide Region

ScanSAR Huge Region

Azimuth resolution Range resolution Azimuth frame extent Range swath extent

~1 m ~1 m ~11 km ~11 km

PRF

3,148.1 Hz to 4,116.7 Hz 70 to 80 ms

~3 m ~3 m ~40 km ~40 km (acq. time ~6.5 s) 2,905.9 Hz to 3,874.5 Hz 35 and 40 ms

~10 m ~10 m ~30 km 30 km (acq. time ~5 s) 2,905.9 Hz to 3,632.4 Hz 30 ms

~30 m ~30 m ~100 km ~100 km (acq. time ~15 s) 2,905.9 Hz to 3,632.4 Hz 30 to 40 ms

~100 m ~100 m ~200 km ~200 km (acq. time ~30 s) 2,905.9 Hz to 3,632.4 Hz 30 to 40 ms

185.2 MHz (187.5 MHz sampling rate)

65.64 MHz (82.50 MHz sampling rate)

14.77 MHz (18.75 MHz sampling rate)

400.0 MHz (187.5 MHz sampling rate)

138.60 MHz (176.25 MHz sampling rate)

38.37 MHz (48.75 MHz sampling rate)

32.74 MHz (41.25 MHz sampling rate) 86.34 MHz (108.75 MHz sampling rate)

8.86 MHz (11.25 MHz sampling rate) 23.74 MHz (30.0 MHz sampling rate)

Linear FM chirp duration LFM chirp bandwidth – minimum

LFM chirp bandwidth – maximum

10.2.6 Italian COSMO/SkyMed System The ‘‘Constellation of Small Satellites for Mediterranean Basin Observation’’ (COSMO)-SkyMed satellites were launched between 2007 and 2010 from Vandenberg AFB, California. All four COSMO-SkyMed satellites are in a single near-polar, sunsynchronous orbital plane at an altitude of approximately 620 km. Data from these satellites support both security-related reconnaissance and civil remote sensing activities. The COSMO-SkyMed SAR operates at a frequency of 9.6 GHz (X-band), which is well suited to the reconnaissance mission. An active phased array antenna is used. The antenna dimensions are 5.7 m by 1.4 m, organized into five panels. The antenna is divided into 40 tiles, each with 32 T/R modules, for a total of 1,280 T/R modules, with each T/R module serving multiple radiating elements. Parameters for the COSMO-SkyMed modes are summarized in Table 10-3. In the enhanced spotlight mode, electronic beam steering increases the time on target to improve the cross-range resolution. The PingPong strip map mode employs alternating transmit and receive polarizations. ScanSAR is also supported [22–24].

10.2.7 German SAR Programs SAR-Lupe. A constellation of five German SAR-Lupe reconnaissance satellites was launched between December 2006 and July 2008 from Plesetsk, Russia. These five satellites are in three orbital planes. The SAR-Lupe radar operates in X-band with a fixed 3.3 m by 2.7 m parabolic antenna. This contrasts with the active phased array antennas used on many other recent spaceborne SARs. In the strip map mode with 1-m resolution, SAR-Lupe can image a 60 km by 8 km area. The spotlight mode is implemented by rotating the satellite bus in the along-track direction, using reaction wheels to

10.3

Orbits

enable the reflector antenna to dwell on the area of interest. This increases the integration time, enabling a resolution of less than 1 m to be achieved over a 5.5 km by 5.5 km image area. TerraSAR-X. TerraSAR-X is a German X-band SAR satellite for civil remote sensing that was launched in 2007. TerraSAR-X will be described in detail in Section 10.6 as a design example of a spaceborne SAR. TanDEM-X. Launched in 2010, TanDEM-X is nearly identical to TerraSAR-X and provides supplemental capabilities for making digital elevation measurements [25]. TerraSAR-X and TanDEM-X operate in formation in essentially the same orbit. To support interferometric SAR modes, a cold gas propulsion system on TanDEM-X permits maneuvering to achieve 200-m to 10,000-m cross-track baselines, with along-track baselines from zero to several hundred meters [26].

10.3

ORBITS

Some of the characteristics of satellite orbits as they pertain to radar remote sensing satellites are discussed briefly in this section. Although this discussion applies to satellites in Earth orbit, the same principles and equations apply to spacecraft orbiting other planets when the appropriate planet radius, planet mass, and/or gravitational acceleration are substituted.

10.3.1 Definitions Before proceeding further, some definitions are in order: apogee or apoapsis: The highest point in a satellite orbit, i.e., the point where the satellite height (or orbit radius) is maximum. The term ‘‘apogee’’ refers specifically to Earth orbits, with the ‘‘gee’’ coming from ‘‘geo.’’ The more general term ‘‘apoapsis’’ refers to the highest point in an orbit around any celestial body. argument of latitude: Angular position of the satellite in its orbit as measured from the ascending node (defined later). The argument of latitude is equal to the sum of the argument of the perigee/periapsis and the true anomaly, which are also defined later [27, 28]. Note that the argument of latitude does not correspond to the satellite’s latitude in Earth coordinates. As the satellite moves around its orbit, the argument of latitude takes on all values from 0 to 2p, regardless of the orbit inclination. argument of the perigee/periapsis: Angle between the ascending node and the perigee or periapsis. ascending node: The point at which the satellite crosses Earth’s equatorial plane from south to north. descending node: The point at which the satellite crosses Earth’s equatorial plane from north to south. inclination: Angle between the satellite orbital plane and Earth’s equatorial plane at the south-to-north equator plane crossing (ascending node). The orbit inclination can range from 0 to 180 and is always positive.

451

452

CHAPTER

10

Space-Based SAR for Remote Sensing

node: A point at which the satellite crosses Earth’s equatorial plane. node line: Intersection of the orbital plane and Earth’s equatorial plane. polar orbit: Strictly speaking, an orbit whose inclination is exactly 90 . The term ‘‘polar orbit’’ is often used loosely to refer to near-polar orbits with inclinations of roughly 70 to 110 . perigee or periapsis: The lowest point in a satellite orbit, i.e., the point where the satellite height (or orbit radius) is minimum. The term ‘‘perigee’’ applies specifically to Earth orbits. The more general term ‘‘periapsis’’ refers to the lowest point in an orbit around any celestial body. prograde orbit: An orbit in which a satellite has a velocity component in the direction of Earth’s rotation. Such orbits have inclinations between 0 and 90 . Many satellites are launched into prograde orbits to take advantage of Earth’s rotational velocity at launch, permitting a larger payload to be delivered to orbit with a given amount of fuel. Other factors can preclude the use of prograde orbits for some missions, however. retrograde orbit: An orbit in which a satellite has a velocity component opposite to the direction of Earth’s rotation. Such orbits have inclinations between 90 and 180 . Many remote sensing satellites are placed into orbits that are near-polar and slightly retrograde (‘‘sun-synchronous’’ orbits, to be discussed later). true anomaly: The angular position of a satellite in its orbit as measured in the direction of satellite motion from the perigee/periapsis. The angle’s vertex is Earth’s center, which is also one of the two foci of the elliptical orbit. vernal equinox direction: The direction defined by a line originating at the center of Earth and passing through the center of the sun at the time of the vernal equinox. This line points toward a specific direction in space beyond the sun, which slowly changes over the 26,000-year precession cycle of Earth’s rotational axis. Although the vernal equinox direction is currently in the constellation Pisces, it was in Aries when observations began over 2,000 years ago. It is, therefore, still referred to as the ‘‘first point of Aries’’ or sometimes simply ‘‘Aries.’’

10.3.2 Circular Orbits Circular orbits are a limiting case of elliptical orbits. Since many remote sensing applications prefer a circular orbit, this type of orbit will be discussed first. Although no satellite orbit is perfectly circular, many are sufficiently close to permit a circular orbit approximation to be used for top-level design calculations and trade-offs. In a circular orbit, the speed of a satellite along its orbital path is rffiffiffiffiffiffiffi m (10.1) vsat ¼ rsat where rsat is the radius of the orbit from the center of Earth in meters and m is a constant defined by m ¼ GðMe þ msat Þ GMe ¼ gs Re 2

(10.2)

10.3

Orbits

where G ¼ universal gravitational constant (6.67 10–11 nt – m2/kg2) Me ¼ mass of the Earth (5.983 1024 kg) msat ¼ mass of the satellite (kg) gs ¼ acceleration of gravity at Earth’s surface (9.81 m/s2) Re ¼ radius of Earth (~6.370 106 m, assumed spherical) The value of m is 3.986 1014 m3/s2. As an example, at an altitude of 800 km, the orbital velocity in a circular orbit is 7.456 km/s. The orbital period is defined as the interval between successive passages of a reference point in the orbit, such as the ascending node. The satellite orbital period (in seconds) is sffiffiffiffiffiffiffiffi 2p rsat rsat 3 P¼ ¼ 2p (10.3) vsat m Conversely, the orbit radius required for a specified orbital period is rsat

2 1=3 mP ¼ 4p2

(10.4)

A satellite in an 800-km-high circular orbit would have an orbital period of 6,042 seconds, or 100.7 minutes. Earth is not a perfect sphere, but rather is an oblate spheroid. It has an equatorial radius of 6,378.1 km and a polar radius of 6,356.8 km, making the radius at the poles 21.3 km less than the radius at the equator [29]. There are additional irregularities in Earth’s shape that can be defined in terms of higher-order spherical harmonics, which are beyond the scope of the present discussion. Earth’s oblateness imparts a torque to a satellite, causing its orbital plane to precess slowly about Earth’s rotational axis. The rate of precession is given by dW 3 pffiffiffiffi Re 3 ¼ J 2 gs cos i dt 2 rsat 7=2

(10.5a)

dW 3 pffiffiffi Re 2 cos i ¼ J2 m dt 2 rsat 7=2

(10.5b)

or, equivalently,

where i is the orbit inclination and W is the right ascension of the ascending node (i.e., the angular difference between the node line and the vernal equinox direction) [30]. (Right ascension is analogous to longitude in Earth-referenced celestial coordinates. It should not be confused with ‘‘celestial longitude,’’ which is defined in sun-referenced, or ‘‘ecliptic,’’ coordinates.) J2 is a coefficient of the second zonal harmonic of the geopotential field. (Zonal harmonics are rotationally symmetric deviations in Earth’s shape from a perfect sphere.) The value of the J2 coefficient is approximately

453

454

CHAPTER

10

Space-Based SAR for Remote Sensing

1.08228 103. All higher-order spherical harmonic coefficients for Earth are at least three orders of magnitude smaller than this value and can generally be neglected. It is evident by inspection that when the inclination is 90 , the orbit plane precession rate will be zero with respect to the fixed stars. Of greater interest is the case where the orbit plane precesses by exactly one full rotation per year. This is termed a ‘‘sun-synchronous’’ orbit because the ascending or descending passes over a given point on Earth’s surface occur at approximately the same local solar time each day. (The word ‘‘approximately’’ is used here because there may not be an integer number of orbits per day and because the satellite orbit precesses at a constant rate while Earth moves at a variable rate in its elliptical orbit around the sun.) Two types of sun-synchronous orbits are of interest in radar remote sensing. First, some sun-synchronous orbits are designed to ensure that the satellite will pass over a given point on Earth’s surface near a particular local time each day, such as 10:00 a.m. Such orbits are often used by remote sensing satellites employing EO/IR sensors to enable their data to be collected with consistent solar illumination. A satellite carrying both radar and EO/IR sensors may be placed in this type of sun-synchronous orbit for the benefit of the optical sensors. In the second type of sun-synchronous orbit, called a dawn-to-dusk or dawn–dusk orbit, the node line remains perpendicular to the solar illumination and the satellite flies approximately along the terminator. For a radar-only satellite, this type of orbit ensures consistency in power production from the solar panels since the satellite is nearly always sunlit except for brief eclipsed periods near the summer solstice. The Canadian RADARSATs use a dawn-to-dusk orbit. Equation (10.5) infers that the inclination required to make a circular orbit sunsynchronous depends on the orbit radius or satellite height. The required inclination can be found by setting dW 2p radians ¼ dt 31;556;926 seconds ¼ 1:99106 107 radians=second ¼ 0:985647 =day where 31,556,926 seconds represents one solar year (365.242199 days), and solving Equation (10.5) for i yields 7=2 dW 2 rsat i ¼ cos pffiffiffiffi dt 3J2 gs Re 3

(10.6a)

7=2 dW 2 rsat i ¼ cos pffiffiffi dt 3J2 m Re 2

(10.6b)

1

or 1

For typical orbits used by remote sensing satellites, the inclination required for a sunsynchronous orbit is on the order of 97 to 103 , or slightly retrograde from a true polar orbit. Figure 10-11 plots the inclination required for a sun-synchronous circular orbit as a function of the orbit altitude together with the period of a satellite at that altitude (regardless of whether or not the orbit is sun-synchronous).

10.3

Orbits

130

FIGURE 10-11 ¢ Period and Inclination Required for a SunSynchronous Circular Orbit.

Inclination (degrees) or Period (minutes)

125 120 115

Period

110 105 100

Inclination

95 90 85 0

500

1,000

1,500

2,000

Satellite Altitude Above Mean Earth Radius (km)

10.3.3 Elliptical Orbits When the orbit is not sufficiently circular, equations for an elliptical orbit must be used. An ellipse is traditionally defined in terms of its semimajor axis a (i.e., half of the major axis) and its eccentricity e. The semimajor axis is a¼

rsatðminÞ þ rsatðmaxÞ hmin þ hmax ¼ Re þ 2 2

(10.7)

where rsat(min) ¼ orbit radius at perigee rsat(max) ¼ orbit radius at apogee hmin ¼ satellite height above mean Earth radius at perigee hmax ¼ satellite height above mean Earth radius at apogee The orbit eccentricity is e¼1

rsatðminÞ a

455

(10.8)

and is zero for a circular orbit. Using these variables, the satellite radius at any point in the orbit is given by að1 e2 Þ rsat ¼ (10.9) 1 þ e cosq where q is the true anomaly (i.e., the angular coordinate of the satellite in its orbit as measured from the perigee) [30]. The satellite velocity in an elliptical orbit varies with the satellite radius [31] and is given by sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 1 (10.10) v¼ m rsat a

456

CHAPTER

10

Space-Based SAR for Remote Sensing

The orbital period (in seconds) is sffiffiffiffiffi a3 P ¼ 2p m and the flight path angle with respect to a local horizontal plane [31] is e sin q 1 gsat ¼ tan 1 þ e cos q

(10.11)

(10.12)

The rate of precession of the orbital plane [1] is ! dW 3 pffiffiffiffi Re 3 cos i ¼ J2 gs dt 2 a7=2 ð1 e2 Þ2

(10.13a)

! dW 3 pffiffiffi Re 2 cos i ¼ J2 m dt 2 a7=2 ð1 e2 Þ2

(10.13b)

or, alternatively,

These expressions reduce to those of Equation (10.5) when e ¼ 0. As with circular orbits, the precession rate of the orbital plane is zero when the inclination is 90 . In addition to orbit plane precession, the angle w between the perigee and the ascending node (the ‘‘argument of the perigee’’) can also shift over time [1], as governed by ! dw 3 pffiffiffiffi Re 3 2 1 5 cos i (10.14a) ¼ J 2 gs dt 4 a7=2 ð1 e2 Þ2 or ! dw 3 pffiffiffi Re 2 1 5 cos2 i ¼ J2 m 2 dt 4 a7=2 ð1 e2 Þ

(10.14b)

This angular rate is zero when the inclination is equal to a ‘‘critical angle’’ ic that can be found by setting 1 5cos2 ic ¼ 0 and solving for ic yielding 1

ic ¼ cos

1 pffiffiffi ¼ 63:4 5

(10.15)

independent of the orbit altitude or eccentricity. An orbit with this inclination would be desirable if the perigee must remain over a specified latitude on Earth, such as when the maximum SNR is desired. Other mission requirements may dictate an orbital inclination different from ic, however. If that is the case, the satellite’s orbit should be as nearly circular as possible to provide uniform SNR performance over the entire orbit.

10.4

Design Considerations for the Spaceborne SAR

10.3.4 Target Doppler Shift Unlike the case of airborne SAR, the Doppler shift of a target or clutter patch observed by a spaceborne radar includes a contribution from Earth’s rotation. The net two-way Doppler shift seen by a satellite in a circular orbit is given [32] by 2vsat we ½ecos bL sin i tan a þ cos i sin qL cos a 1 (10.16) fd ¼ l wsat where qL ¼ look angle from nadir a ¼ azimuth squint angle (about the yaw axis) between the velocity vector and the target azimuth we ¼ Earth’s angular rotation rate wsat ¼ satellite angular rate ¼ 2p/P radians/sec for a circular orbit e ¼ –1 if the radar is left-looking; þ1 if right-looking bL ¼ argument of latitude (angular position of the satellite in its orbit as measured from the ascending node) i ¼ orbit plane inclination The term in brackets is due to Earth’s rotation. If we is zero, the term in brackets vanishes and the equation reduces to the form used for airborne radars in which the radar moves in an Earth-referenced frame. Derivations for both circular and elliptical orbits are presented in Appendix B of Curlander and McDonough [33]. Based on Equation (10.16), the azimuth squint angle resulting in zero Doppler shift is

N cos i 1 (10.17) e a0 ¼ tan cos bL sin i where N is the number of orbits per day (N ¼ wsat/we). This angle typically deviates by plus or minus a few degrees from 90 (side-looking) and varies sinusoidally during the orbit. The beam should ideally be pointed in this direction in strip map modes to center the surface returns at DC (zero Doppler). Some of the newer satellites have this capability, which reduces throughput requirements in the ground-based processor.

10.4 DESIGN CONSIDERATIONS FOR THE SPACEBORNE SAR This section addresses design issues and trade-offs that are unique to spaceborne SARs. More general aspects of SAR design are covered in many excellent references. In particular, Curlander and McDonough [33] thoroughly cover all issues associated with strip map SARs, both spaceborne and airborne. Carrara et al. [34] is a detailed and practical reference on spotlight SAR imaging from airborne platforms. Jakowatz et al. [35] and Cumming and Wong [36] provide greater detail on SAR processing. The Cumming and Wong book in particular emphasizes processing of SAR data from spaceborne radars.

457

458

CHAPTER

10

Space-Based SAR for Remote Sensing

10.4.1 Propagation Phenomena 10.4.1.1 Ionospheric Effects Faraday Rotation. When a linearly polarized wave passes through the ionosphere, the plane of polarization rotates due to a phenomenon called Faraday rotation. This occurs because the phase delay through the ionosphere is different for different polarization components with respect to Earth’s magnetic field. The amount of polarization rotation depends on the total electron content (TEC) along the path and the geometric relationship between the propagating wave and Earth’s magnetic field. Free electrons are present in the ionosphere as a result of ionization of upper atmospheric atoms by solar ultraviolet radiation. The degree of ionization varies with the time of day, season, the 11-year sunspot cycle, short-term solar activity (flares, etc.), and ionospheric dynamics. Thus, most ionospheric phenomena, including Faraday rotation, are highly variable. The amount of Faraday rotation is given approximately [37] by WP ¼

2:365 104 TEC B0 cos qB f 2 cos j

(10.18)

where WP ¼ rotation of the polarization plane (radians) TEC ¼ total electron content in a vertical column through the ionosphere with a cross-section of 1 m2 (electrons/m2) B0 ¼ magnetic field strength ¼ 0.5 104 teslas ( ¼ 0.5 gauss) qB ¼ angle between the direction of propagation and the magnetic field f ¼ radar frequency (Hz) j ¼ angle of the propagation path in the ionosphere from nadir (approximately equal to the off-nadir look angle qL) The primary effect of Faraday rotation on linearly polarized radar returns is a loss in the return power due to the polarization mismatch between the radar returns and the receiving antenna. This loss can degrade the data quality in a single-polarization radar and can alter polarization measurements in dual-polarization radars. Equation (10.18) shows that the magnitude of the Faraday rotation is proportional to 2 1/f . At low frequencies (

2 RsðmaxÞ RsðminÞ c

(10.30)

2DRs c

(10.31)

or, more compactly, Tr >

where DRs ¼ Rs(max) – Rs(min) [2]. Equivalently, the PRF (fr) is constrained to be fr

2kaz vsat L

(10.45)

This expression relates the minimum PRF to the antenna length in the along-track direction, an important trade-off in SAR design. Substituting L ¼ 2dCR from Equation (10.39) produces fr >

kaz vsat dCR

(10.46)

which relates the minimum PRF to the required cross-range resolution, which is limited by the antenna length in the strip map mode. Clearly there is a complex interplay between the cross-range resolution requirement, the along-track antenna length, and the minimum PRF. When the relationship between the PRF and the elevation beamwidth is also considered, it is often difficult to arrive at a completely optimal design, although suboptimal designs have been flown very successfully. These issues are addressed in greater detail in [46].

10.4.5.3 Minimum Aperture Area Requirement for SAR Solving Equation (10.35) for H yields the minimum antenna height for a given PRF based on range ambiguity considerations, i.e., H>

2fr Rs kel l tan qi c

(10.47)

Similarly, solving Equation (10.45) for L provides the minimum antenna length for a given PRF based on azimuth ambiguity considerations, or L>

2kaz vsat fr

(10.48)

10.4

Design Considerations for the Spaceborne SAR

When the product of Equations (10.47) and (10.48) is taken, the PRF cancels and the result is a minimum aperture area requirement for a given frequency, satellite velocity, and radar-target geometry: A ¼ HL >

4Rs lkaz kel vsat tanqi c

(10.49)

This result is also given in [32] and [46]. It should be noted that this is not a hard limit. Through careful design, spaceborne SARs violating this criterion have flown successfully, realizing a cost savings through a reduced aperture area.

10.4.6 Block Adaptive Quantization (BAQ) Most of the signal and image formation processing in spaceborne remote sensing radars is usually done on the ground. Also, many spaceborne radars have a limited downlink capacity. Since complex I/Q data are fairly noiselike, conventional data compression algorithms provide little benefit since they generally try to exploit correlations and redundancies in the data. One data compression approach that has been used fairly successfully involves reducing the number of ADC bits per sample that are transmitted to the ground. For example, an 8-bit quantizer may be used to digitize the IF or baseband return signals, but only a subset of the eight bits is sent to the ground. The specific bit range selected depends on the characteristics of the data, so this subset can vary over time. This approach to data compression is called block adaptive quantization (BAQ) or block floating point quantization (BFPQ). Block adaptive quantization is widely used in spaceborne radars, as it helps to maximize the utility of the spaceborne assets when the system is data rate limited. BAQ takes advantage of the fact that the dynamic range of the raw samples is limited within restricted portions of the raw data set for an image. This is especially true when pulse compression is used (which is almost always the case) because the uncompressed returns from various scatterers in the scene overlap in time, reducing the overall variability in the samples. BAQ is usually done with buffered blocks of data rather than being applied ‘‘on-the-fly’’ to a continuous data stream. For BAQ to be effective, the variance of the samples within each block should be small compared with the variance across all blocks in the data set. The criteria for selecting an appropriate block size [33] are: 1. The block size should be >50–100 samples to ensure Gaussian statistics in the presence of speckle. 2. The block size should have a limited range extent to avoid power variation due to range (R–3 for distributed surface clutter) and elevation beam pattern roll-off. The variation due to these effects should be kept to 2,000 ops/pixel per pol. channel

PWF

100 Ops/pixel

30 Ops/pixel

SAR Images With Interference

Screener 200–500 ops/ pixel

RFI Rejection

Detections & FAs

3 Polarization Images

PWF Image

Extract Chip of Each ROI

Feature Extraction

1,000 Ops/ROI

15,000 Ops/ROI Chip of ROI

Classification 1,000 ops/ROI

FIGURE 14-40 ¢ FOPEN SAR Target Detection and Characterization Processing Flow. [Source: MIT Lincoln Laboratory [45]]

678

CHAPTER

14

Foliage-Penetrating Radar

ratio and the subsequent ability to nominate areas of the image that are likely to contain manmade targets. Target detection is accomplished through an area CFAR on the image, providing a nomination of regions of interest (ROIs) for subsequent feature examination. These regions are then delineated as small areas of target and background clutter, called chips, and passed on for subsequent feature extraction. These ‘‘chips’’ will be examined finally for features that include size, shape, and polarization characteristics for final classification as a manmade target versus terrain artifact.

14.7.2 Polarization Whitening Speckle has been recognized to be the cause of significant false alarms in SAR images [41]. As a result, the consideration of polarization diversity as a technique to reduce this speckle has been examined extensively, primarily for microwave frequency radars. The analysis of polarimetric returns from manmade targets and clutter provided verification of the benefit of using the independent degrees of signal polarization characteristics. Based on these results and the known spiky characteristics of clutter, it was straightforward to investigate applications to assist in the reduction of false alarms in foliagepenetrating SAR. The P-3 UWB SAR was specifically built to collect fully polarimetric clutter and target characteristics over a wide range of geographic and forested regions. It was shown in Section 14.4 that the foliage clutter is non-Gaussian. With a Gaussian clutter model, each resolution cell of the SAR image will be spatially homogeneous and have the same average polarimetric power. However, with the foliage lognormal clutter model, this characteristic cannot be assumed. In fact, there is a significant inhomogeneity of the clutter distribution over a typical SAR scene. A polarimetric FOPEN SAR system typically collects three polarization components by using two receiver channels. On successive pulses, horizontal and vertical transmit pulses illuminate the scene, and the co-polarization and cross-polarization returns are recorded. Because the two cross-polarization channels HV and VH have been observed and analyzed as being reciprocal, only three channels are typically recorded: HH, HV, and VV. Polarization whitening is the process that combines the three input channels into a complex vector in order to equalize the energy in three polarization vector quantities. The polarization measurements of the signal return are collected into a complex vector [43]: 2 3 2 3 HHI þ jHHQ HH Y ¼ 4 HV 5 ¼ 4 HVI þ jHVQ 5 (14.37) VVI þ jVVQ VV The vector Y is assumed to be the product of a complex Gaussian vector X (representing the speckle) and a spatially varying texture variable g: pffiffiffi Y ¼ gX (14.38) The probability density function of the complex speckle vector X is defined by f ðX Þ ¼

1 p3 jSj

expðX S1 X Þ

(14.39)

14.7

Target Detection and Characterization

where S is the polarization covariance matrix. It is common to express the clutter covariance matrices in terms of normalized linear-polarization bases in the form: 0 pffiffiffi 1 1 0 r g e 0 A S ¼ sHH @ 0 (14.40) pffiffiffi r g 0 g In Equation (14.40), the parameter g is the ratio of the square of the expectation value of the VV to that of the HH return: g¼

EðjVV j2 Þ EðjHH j2 Þ

(14.41)

Similarly, e is the ratio of the HV cross-polarization to the HH return: e¼

EðjHV j2 Þ EðjHH j2 Þ

(14.42)

Finally, r is the cross-correlation coefficient between the HH and VV returns: r¼

EðHH VV Þ

(14.43)

½EðjHHj2 ÞEðjVV j2 Þ1=2

From the polarimetric measurements at each pixel, the objective is to construct a new image that minimizes the variance of speckle between the three channels. This is carried out by a transformation of the complex vector Y in the quadratic form: y ¼ Y y A Y ¼ g X yA X

(14.44)

The measure of speckle in the image is expressed as the ratio of the standard deviation of the image pixel intensities to the mean of the intensities (s/m): sy std:devðyÞ ¼ my meanðyÞ

(14.45)

A is the desired weighting matrix that minimizes the (s/m) in each channel. This has been shown to equal the inverse of the polarization covariance matrix. So the polarization vector Y from each pixel in the image is passed through the whitening filter S1/2 to obtain a new image function YW: YW ¼ S1=2 Y ¼

pffiffiffi 1=2 gS X

(14.46)

The whitened vector YW then forms a single SAR image with the requisite reduction in background clutter speckle, given by [41]: 2 6 YW ¼ 4HH;

HV pffiffiffi ; e

pffiffiffi

3

VV r gHH 7 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 5 gð1 jrj2 Þ

(14.47)

679

680

CHAPTER

14

Foliage-Penetrating Radar

The scale factors in Equation (14.47) come from the three ratios of image polarization channel data given by g, e, and r in Equations (14.41) to (14.43). Figure 14-41 shows the three P-3 FOPEN SAR polarimetric input channels from the Grayling, Michigan, collection [43]. It can be clearly seen that the HH and VV channels exhibit a spiky distribution of clutter discretes, representative of a lognormal distribution, as summarized in Section 14.4. To characterize this clutter distribution, a set of 500 clutter chips was extracted from the images, each containing 200 pixels in range and 100 pixels in cross-range. The values of the parameters of the tree clutter polarization covariance matrix were then estimated. For this type of northern latitude forest clutter, the parameters are found to be approximately [42]: sHH ¼ 0:08 e ¼ 0:25 g ¼ 0:5 pffiffiffi r g ¼ j 0:125

(14.48)

When the three channels were combined with the PWF, the clutter distribution of scatterers is modified significantly. This is evidenced in Figure 14-41 by the closer distribution of return amplitudes in the forested area and the reduction in speckle noise in the open field. The SCR for targets under the foliage was also improved from 9 dB to 19 dB. More importantly, the targets suffered no loss of resolution. Thus, it is expected that after PWF processing, the target detection process will be greatly improved.

14.7.3 Target Characterization With efficient image formation and RFI mitigation, the polarimetric FOPEN SAR presents an opportunity to improve the detection of tactical targets hidden under forest cover. However, the reduction of the target signature, due to both the foliage-penetrating FIGURE 14-41 ¢ Polarization Whitening of Grayling P-3 UWB Data. [Source: MIT Lincoln Laboratory [45]]

HH

VV

Polarization Whitening Filter

PWF

HV

14.7

Target Detection and Characterization

losses and the similarity of many natural objects in size to the desired targets, falsealarm rate remains high. Further image processing techniques were developed to discriminate returns that pass the CFAR test. Polarization is one of many techniques developed to determine whether an individual object is manmade or natural. However, there has been an equally strong emphasis on increased image resolution in an attempt to segregate the scattering centers and to reduce the competing clutter volume. Both fine spatial resolution and polarization have been shown to improve target feature characterization [46]. Other strong image processing techniques, as outlined in Table 14-8, were examined to understand the geometric characteristics of tactical objects under the forest canopy, including determining the geometric and polarimetric features of targets versus terrain feature [47, 48]. Several data collections have been conducted to obtain sufficient clutter and target data to test the process of image segmentation and strength of target characterization for both false-alarm reduction and classification of manmade targets. It is not anticipated that FOPEN SAR will provide any useful automatic target recognition due principally to two factors. First, at the long wavelengths of UHF and VHF, there are insufficient pixels on a target to expect a good recognition performance. But more importantly, the few features that exist on manmade and tactical objects will be occluded by the propagation of the signal energy through large trees. These essentially mask many of the pixels and cause amplitude and phase perturbation of the radar returns. So the most that can be expected from FOPEN SAR automatic target detection and characterization (ATD/C) is to place objects in broad classes based on the features that do exist robustly in the measurements.

14.7.4 FOPEN Target Features Four types of features have been employed in classification analyses, as summarized in Table 14-8 [46]: ●

Texture: The spatial variation of the returns in the neighborhood or as part of the area being characterized. The returns for clutter and target within the CFAR box are examined for their particular statistical characteristics. The standard deviation of the TABLE 14-8 Objects [47]

¢

Image Processing Techniques for Detecting Manmade

Feature Types

Algorithm

Textural

Standard deviation Fractal dimension Ranked fill ration Mass Diameter Normalized square rotational inertia Peak CFAR Mean CFAR Percent bright CFAR Percent pure (odd or even) Percent pure even Percent bright even

Size

Contrast features

Polarimetric features (Fully polarimetric only)

681

682

CHAPTER

14

Foliage-Penetrating Radar

FIGURE 14-42 ¢ Estimating Size and Shape Features for Target Characterization.

CFAR Clutter Estimate Minor Axis

Major Axis

α

Target Morphological Shape

returns in the area is a measure of the fluctuation of the intensity in an image. The fractal dimension measures the N brightest scatterers in the region and characterizes them in terms of geometric shape. A single bright pixel will have a dimension of 0, a line will have a dimension of 1, and a solid rectangle will have a dimension of 2. Finally, the rank fill feature measures the percentage of total energy contained in the N largest pixels. N is typically chosen as 5. ●

Size: The grouping of strong scatterers and association with an object via a morphological filter for determination of length and width, as illustrated in Figure 14-42. The mass feature is obtained by counting the number of pixels in this morphological shape. The normalized square rotational inertia feature is the second mechanical moment of the shape around its center of mass, normalized by the inertia of an equal mass square.

The center of mass of the object within the image chip can be calculated by the weighted position of each pixel that exceeds the CFAR threshold. Given (xi, yi) as the position of the pixel, the centers of mass Mx and My are calculated by P P xi yi My ¼ i (14.49) Mx ¼ i N N The second moment of the image (i.e., its inertia) is now given in three dimensions, based on the distance between each detected image pixel and the center of mass of the image: Uxx ¼

1X ðxi Mx Þ2 N i

(14.50a)

Uyy ¼

1X ðyi My Þ2 N i

(14.50b)

1X ðxi Mx Þðyi My Þ N i

(14.50c)

Uxy ¼

14.7

Target Detection and Characterization

Calculation of the critical parameters for target discrimination are next determined by estimating the pose (i.e., orientation) of the shape with respect to the x- and y-axes. Based on these measurements, the pose of the object is determined as 1 2Uxy a ¼ tan1 (14.51) 2 Uxx Uyy The coordinate system can now be transformed from (x, y) to (x0 , y0 ) by an a-rotation in the image plane. The lengths of the major and minor axes are next determined by centroiding the locations of the N brightest pixels: rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1X 0 2 UMajor ¼ ðxi Þ (14.52a) N rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 X 0 2 (14.52b) ðyi Þ UMinor ¼ NC ●

Contrast: The collection of features obtained from the statistics in the CFAR processing. The ratio and distribution of the scattering center returns in the target shape (shown in Figure 14-42) are compared to the background clutter statistics. The maximum statistic is the greatest intensity pixel in the target distribution. The mean feature is the average of the CFAR target returns within the morphological shape, while the percentage bright feature is the percentage of the pixels within the object that exceeds a CFAR threshold. Polarization: It has been found to provide a very effective method to discriminate between manmade and natural scatterers, as summarized in Section 14.4. Oddbounce statistics are representative of flat-plate, linear (e.g., wires), or trihedral objects. Even-bounce returns are associated with the radar scattering from a dihedral. Thus, the polarimetric features are calculated from a transformation of the HH-VV data collection coordinates to even-bounce and odd-bounce images. These transformations are given by the following equations: jSHH þ SVV j2 2

(14.53)

jSHH SVV j2 þ 2jSHV j2 2

(14.54)

Eeven ¼ Eodd ¼

The percentage-pure feature is the fraction of the pixels within the target shape for which at least a threshold of the scattered energy falls within either the even-bounce or odd-bounce metric. The percentage-bright-even feature is the fraction of the pixels within the target shape that exceed a threshold in the CFAR image, which are predominantly even-bounce scatterers. Evaluating the pixels in a target chip after image segmentation or designation of ROIs provides additional quantitative scoring to the even- and odd-bounce features for additional image characterization. First, the SHH/SHV and SHH/SVV ratios are calculated for each pixel. Then the average and standard deviation of the measurements are

683

684

CHAPTER

14

Foliage-Penetrating Radar

calculated for the number of pixels in the ROI. The six additional polarimetric features that need to be calculated are jSHH j jSHH j P4 ¼ Average P1 ¼ Average jSHV j jSVV j jSHH j jSHH j P2 ¼ Std:Dev: (14.55) P5 ¼ Std:Dev: jSHV j jSVV j AveragejSHH j AveragejSHH j P6 ¼ P3 ¼ AveragejSHV j AveragejSVV j The final step in the target discrimination process combines the appropriate features from the aforementioned extensive list into a single discrimination statistic. The features can then be collected and examined in terms of a quadratic distance metric to identify classes of objects for either classification or elimination as false alarms. This discrimination statistic is calculated as a quadratic distance metric: X 1 1 b ÞT bÞ dT ðX Þ ¼ ðX M (14.56) ðX M N

^

where N ¼ number of features used for discrimination M ¼ estimate of the mean target vector class S ¼ estimate of the standard deviation target vector class dT (X ) ¼ statistical distance of the unknown object from a class The estimates of M and S are often collected on targets in the open to obtain a fundamental understanding of the target features. The measured quadratic distance dT(X) will be small for targets that are close to the correct class. For natural clutter, it is anticipated that the quadratic measure will be large for scattering that does not exhibit the dominant even- or odd-bounce characteristics of manmade objects. The challenge in FOPEN ATD/C has always been the occlusion (i.e., shadowing) of the targets by dense foliage. If the target’s statistics are collected for manmade objects in the open, there will be an inherent error in the mean and standard deviation of the target classes. These errors will be larger if the target SCR is low, or if there are very large natural objects in the direct path. As a result, any collection of data to test ATD/C capabilities need to be gathered in a variety of foliage conditions, including types of forests, terrain slope, and radar incident angles.

14.8

SUMMARY

Foliage-penetrating radar has been in development for over 40 years. The early systems provided an important insight into the scattering characteristics of forest and jungle environments, as well as the requirements for coherent waveforms and signal processing. The attenuation of propagation of radar through foliage was not as severe as originally expected, but only as long as the propagation occurred at a grazing angle (to the top of the foliage) above approximately 20 degrees. However, the motion of foliage in

14.9

References

any wind conditions would mask the motion of moving targets and obscure any target features. As a consequence, the further development of FOPEN radar needed to wait for advances in compact VHF- and UHF-band components, as well as high-performance computing. FOPEN SAR requires the use of UWB waveforms and a significant percentage of the spectrum. Two factors became an early limitation on FOPEN resurgence in the late 1990s. First, the large number of VHF and UHF transmitters for radio, television, and communications was a constant source of interference in the formation of high-quality FOPEN images. Second, the regulations on transmissions outside of the radar bands required significant advances in the low probability of intercept frequency operation. These conditions still exist, and acceptance of any new system requires comprehensive testing and verification that no disruption to emergency and safety support systems exists. In order for efficient target detection, FOPEN SAR signal processing must consider the removal of RFI as well as maintaining high-quality images. Polarization features are shown to be important in the basic detection process, as well as discriminating manmade targets from cultural objects. Future systems will rely on the continued maturing of image processing as well as the steady improvement in real-time processing.

14.9

REFERENCES

[1] Gordon G. A. and Holt E., ‘‘An Estimate of the HF/VHF Surface-Wave Communications Wave Reaches in the West German Forest Environment,’’ Defense Nuclear Agency Report DNA-TR-82-07, January 1982, pp. 12–17. [2] Surgent L. V. Jr., ‘‘Foliage Penetration Radar: History and Developed Technology,’’ U.S. Land Warfare Laboratory Report AD/A000805, July 1974 (Publically released). [3] Centofanti J. J., ‘‘Synthetic Aperture Dual Frequency Radar (SADFRAD) – A Unique Airborne Sensor,’’ AFCRL-70-0676, Air Force Cambridge Research Laboratories, Hanscom Field Bedford, MA, AD515249, 9 December 1970 (Publically released 31 December 1982). [4] Fowler, C. A., ‘‘The Standoff Observation of Enemy Ground Forces: From Project PEEK to JointSTARS,’’ IEEE Systems Magazine, Vol. 12, No. 6, June 1997, pp. 3–17. [5] Entzminger J. N., Fowler C. A., and Kenneally W. J., ‘‘Joint STARS and GMTI: Past, Present, and Future,’’ IEEE Transactions on Aerospace and Electronic Systems, Vol. 35, No. 2, April 1999, pp. 748–761. [6] Bryant T. G., Morse G. B, Bovak L. M, and Henry J. C., ‘‘Tactical Radars for Ground Surveillance,’’ MIT Lincoln Laboratory Journal, Vol. 12, No. 2, Lexington, MA, 2000, p. 342. [7] Picture of Camp Sentinel Radar Antenna courtesy of Lee Moyer Technology Services Company, Silver Springs, MD. [8] Photographs and details of M-FOPEN radar provided by Mr. James Rodems, former Syracuse University Research Company division manager. [9] Davis M. E., ‘‘Technical Challenges in Ultra-Wideband Radar Development for Terrain Mapping,’’ Proc. IGARSS, 1998, Seattle, WA. [10] Carrara W. G., Goodman R. S., and Majewski R. M., Spotlight Synthetic Aperture Radar Signal Processing Algorithms, Artech House, Boston, MA, 1995, Chapter 2.

685

686

CHAPTER

14

Foliage-Penetrating Radar

[11] Vickers R. S., ‘‘Ultra Wideband Radar – Potential and Limitations,’’ Proc. 1991 IEEE MTTS Conference, June 1991, pp. 371–374. [12] Hellsten H., Ulander L. M. H., Gustavsson A., and Larsson B., ‘‘Development of VHF CARABAS II SAR,’’ Proc. Radar Sensor Technology, SPIE, Vol. 2747, Orlando, FL, 8–9 April 1996. [13] Vandenberg N., Sheen D. R., Shackman S., Wiseman D., ‘‘P-3 Ultra Wide Band SAR: System Applications to Foliage Penetration,’’ SPIE, Vol. 2757, Orlando, FL April 1996, p. 130. [14] Hensley S., Chapin E., Freedman A., Le C., Madsen S., Michel T., Rodriguez E., Siqueira P., and Wheeler K., ‘‘First P Band Results Using the GeoSAR Mapping System,’’ Proc. 2001 IEEE Radar Conference, Atlanta, GA, May 2001, p. 126. [15] Ressler M., Happ L., Nguyen L., Tuan T., and Bennett M., ‘‘The Army Research Laboratory Ultra-Wide Band Testbed Radars,’’ Proc. IEEE International Radar Conference, Arlington, VA, May 1995, pp. 686–691. [16] Adams G. F., Ausherman D. A., Crippen S. L., Sos G. T., and Williams B. P., Heidelbach R., ‘‘The ERIM Interferometric SAR: IFSARE,’’ Proc 1996 National Radar Conference, Ann Arbor, MI, 13–16 May 1996, pp. 249–254. [17] Courtesy of M. Toups, MIT Lincoln Laboratory, Lexington, MA, June 1997. [18] Fleischman J., Toups M. F., and Ayasli S., ‘‘Summary of Results from a Foliage Penetration Experiment with a Three Frequency Polarimetric SAR,’’ Proc. SPIE, Vol. 1693 Surveillance Technologies II, April 1992, p. 151. [19] Bessette L. A. and Ayasli S., ‘‘Ultra Wideband P-3 and CARABAS II Foliage Attenuation and Backscatter Analysis,’’ Proc. 2001 IEEE Radar Conference, Atlanta, GA, May 2001, p. 357. [20] Bessette L. A., Crooks S. M., and Ayasli S., ‘‘P-3 Ultra Wideband SAR Grayling, Michigan, Target and Clutter Phenomenology,’’ Proc. 1999 IEEE Radar Conference, Boston, MA, May 1999, p. 125. [21] Moyer L. R., private communication, Technology Services Corporation, Silver Springs, MD, February 2010. [22] Davis M. E., Tomlinson P. R, and Maloney R. P., ‘‘Technical Challenges in Ultra- Wideband Radar Development for Target Detection and Terrain Mapping,’’ Proc. 1999 IEEE Radar Conference, Boston, MA, pp. 1–6. [23] Hellsten H., ‘‘CARABAS – a UWB Low Frequency SAR,’’ IEEE MTT-S International, Vol. 3, 1–5 June 1992, pp. 1495–1498. [24] Vickers R., Gonzalez V. H., and Ficklin R. W., ‘‘Results from a VHF Impulse Synthetic Aperture Radar,’’ Proc. SPIE, Vol. 1631, Bellingham, WA, 1992, pp. 219–226. [25] Yegulalp A. R., ‘‘Fast Backprojection Algorithm for Synthetic Aperture Radar,’’ Proc. 1999 IEEE Radar Conference, Boston, MA, April 1999, pp. 60–65. [26] Hunter A. J., Hayes M. P., and Gough P. T., ‘‘A Comparison of Fast Factorised BackProjection and Wavenumber Algorithms for SAS Image Reconstruction,’’ Proc. of the World Congress on Ultrasonics, Paris, France, September 2003. [27] Soumekh M., Synthetic Aperture Radar Signal Processing with MATLAB Algorithms, Wiley, New York, 1999. [28] Miller T., McCorkle J., and Potter L., ‘‘ Near Least Square Radio Frequency Interference Suppression,’’ Proc. SPIE, Vol. 2487, Orlando, FL, April 1995, pp. 72–83.

14.9

References

[29] FCC Part 15 Regulations July 10, 2008, from Federal Communications Commission Internet Site, http://www.fcc.gov/oet/info/rules/part15/PART15_07-10-08.pdf. [30] Lynch D. Jr., Introduction to RF Stealth, SciTech Publishing, Raleigh, NC, 2004, Chapter 2: Interceptability Parameters and Analysis. [31] Davis M. E., ‘‘Waveforms for Ultra Wideband FOPEN SAR,’’ Proc. 2010 International Radar Conference, Washington, DC, May 2010. [32] Goodman R., Tummala S., and Carrara W., ‘‘Issues in Ultra-Wideband, Widebeam SAR Image Formation,’’ Proc. IEEE 1995 International Radar Conference, Arlington, VA, May 1995, pp. 479–485. [33] Chang W., Cherniakov M., Li X., and Li J., ‘‘Performance Analysis of the Notch Filter for RF Interference Suppression in Ultra-Wideband SAR,’’ Proc. 9th International Conference on Signal Processing (ICSP200), 2008, pp. 2446–2451. [34] Le C. T. C. and Hensley S., ‘‘Removal of RFI in Wideband Radars,’’ Proc. 1998 IGARSS Symposium, Seattle, WA, July 1998, p. 2032. [35] Golden A., Werness S. A., Stuff M., DeGraaf S., and Sullivan R., ‘‘Radio Frequency Interference Removal in a VHF/UHF Deramp SAR,’’ Proc. SPIE, Vol. 2487, Orlando, FL, April 1995, pp. 84–95. [36] Tufts D. W. and Kumaresan R. ‘‘Estimation of Frequencies of Multiple Sinusoids: Making Linear Prediction Perform Like Maximum Likelihood,’’ Proc. IEEE, Vol. 70, No. 9, September 1982, pp. 975–989. [37] Toups, M., unpublished images provided by MIT Lincoln Laboratory as part of DARPA/ Navy UWB P-3 Collection and Verification Program, 1997. [38] Hensley S. and Madsen S. N., ‘‘Interferometric Radar Waveform Design and the Effective Interferometric Wavelength,’’ Proc. 2007 Waveform Diversity & Design Conference, Kauai, HI, April 2007, pp. 287–291. [39] Le C. T. C., Hensley S., and Chapin E., ‘‘Adaptive Filtering of RFI in Wideband Radars Using the LMS Algorithms. Part I: The TDLMS Adaptive Filter,’’ JPL Technical Symposium, Pasadena, CA, 2002. [40] Hensley S., Le C. T. C., and Gurrola E., ‘‘A Rigorous Expression for the Amount of Interference from a Chirped Waveform Using Ideal Band Pass and Realistic Receiver Models,’’ NASA Jet Propulsion Laboratory GeoSAR Memo (unpublished), September 20, 1999. [41] Novak L. M., Burl M. C., and Irving W. W., ‘‘Optimal Polarimetric Processing for Enhanced Target Detection,’’ IEEE Trans. on Aerospace and Electronic Systems, Vol. 29, No. 1, January 1993, pp. 234–243. [42] Novak, L. M., ‘‘Effects of Various Image Enhancement Techniques on FOPEN Data,’’ Proc. 2001 IEEE Radar Conference, Atlanta, GA, May 2001, p. 87. [43] Nanis J. G., Halversen S. D., Owirka G. J., and Novak L. M., ‘‘Adaptive Filters for Detection of Targets in Foliage,’’ IEEE AESS Magazine, August 1995, p. 34. [44] Schuler D. L., Lee J-S., and De Grandi G., ‘‘Measurement of Topography Using Polarimetric SAR Images,’’ IEEE Trans. on Geoscience and Remote Sensing, Vol. 34, No. 5, September 1996. [45] Davis M. E., Tomlinson P. G., and Malone R. P., ‘‘Technical Challenges in Ultra Wideband Radar Development for Target Detection and Terrain Mapping,’’ Proc. 1999 IEEE Radar Conference, Boston, MA, April 1999, p. 1.

687

688

CHAPTER

14

Foliage-Penetrating Radar

[46] Novak L. M., Halversen S. D., and Owirka G. J., ‘‘Effects of Polarization and Resolution on the Performance of a SAR Automatic Target Recognition System,’’ MIT Lincoln Laboratory Journal, Vol. 8, No. 1, January 1995. [47] MacDonald D., Chang C. F., Roman J., and Koesel R., ‘‘Automatic Target Detection and Cueing for Foliage-Concealed Targets,’’ Proc. SPIE, Vol. 2757, Orlando, FL, April 1996, pp. 152–162. [48] Gorman J. D. and Marble J. A., ‘‘A Low Complexity Multi-Discriminant FOPEN Target Screener,’’ Proc. SPIE, Vol. 3370, Orlando, FL, April 1998, pp. 307–317.

14.10

FURTHER READING

Further details on the development and technology for foliage-penetrating radar can be found in the following: Davis M. E., Foliage Penetration Radar – Detection and Characterization of Objects Under Trees, SciTech Publishing, Raleigh NC, 2011. Several key papers that cover the essential elements of FOPEN phenomenology and radar design include: FOPEN History Surgent L.V. Jr., ‘‘Foliage Penetration Radar: History and Developed Technology,’’ U.S. Land Warfare Laboratory Report AD/A000805, July 1974 (Publically released). Bryant T. G., Morse G. B., Novak L. M., and Henry J. C., ‘‘Tactical Radars for Ground Surveillance,’’ Lincoln Laboratory Journal, Vol. 12, No. 2, Lexington, MA, 2000, p. 342. Phenomenology Bessette L. A. and Ayasli S., ‘‘Ultra Wideband P-3 and CARABAS II Foliage Attenuation and Backscatter Analysis,’’ Proc. 2001 IEEE Radar Conference, Atlanta, GA, May 2001, p. 357. RFI Removal Le C. T. C. and Hensley S., ‘‘Removal of RFI in Wideband Radars,’’ Proc. 1998 IGARSS Symposium, Seattle, WA, July 1998, p. 2032. Golden A., Werness S. A., Stuff M., DeGraaf S., and Sullivan R., ‘‘Radio Frequency Interference Removal in a VHF/UHF Deramp SAR,’’ Proc. SPIE, Vol. 2487, Orlando, FL, April 1995, pp. 84–95. Waveform Processing Carrara W. G., Goodman R. S., and Majewski R. M., Spotlight Synthetic Aperture Radar Signal Processing Algorithms, Artech House, Boston, MA, 1995. Soumekh M., Synthetic Aperture Radar Signal Processing with MATLAB Algorithms, Wiley, New York, 1999.

14.10

Further Reading

Target Detection and Discrimination Novak L. M., Halversen S. D., and Owirka G. J., ‘‘Effects of Polarization and Resolution on the Performance of a SAR Automatic Target Recognition System,’’ MIT Lincoln Laboratory Journal, Vol. 8, No. 1 January 1995. MacDonald D., Chang C. F., Roman J., and Koesel R., ‘‘Automatic Target Detection and Cueing for Foliage-Concealed Targets,’’ Proc. SPIE, Vol. 2757, Orlando, FL, April 1996, pp. 152–162.

689

CHAPTER

Ground-Penetrating Radar

15

Richard Liu, Wei Ren, Huaping Wang and Chen Guo, Department of Electrical and Computer Engineering, University of Houston, Houston, Texas 77004 Chapter Outline 15.1 15.2 15.3 15.4 15.5

Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pulsed Ground-Penetrating Radar System Design . . . GPR System Implementation and Test Results. . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

15.1

....... ....... ....... ....... .......

........ ........ ........ ........ ........

.... .... .... .... ....

691 697 731 746 746

OVERVIEW

Ground-penetrating radar (GPR) is a noninvasive electromagnetic geophysical survey device. GPR uses high-frequency electromagnetic (EM) wave propagation (generally, 10 MHz to 10 GHz) and scattering to image, locating and quantitatively identifying changes in electromagnetic properties in materials, especially in the ground. GPR measurements may be conducted from the surface of the materials, from a small distance (0.5–2 m) above the material surface, in a borehole or between boreholes, and even from an aircraft. GPR offers a high spatial resolution in the subsurface imaging as a geophysical method. Depending on the antennas used, the spatial resolution can reach centimeters or even millimeters under the right conditions. The depth of investigation varies from less than a meter to more than tens of meters, depending on the material properties and antennas used [1]. The areas of application for the GPR are diverse [2–10] and include the detection or mapping of metallic features, which include utility lines, archeological sites, and subsidence in soil. The system has also been used successfully to determine the structural integrity of the ground underneath a road or runway. This results in applications involving the identification of defects within pavements and bridge decks, measuring moisture contents, detecting corrosion of reinforcing steel in concrete [3], and assessing the thickness of pavement layers [4–8]. GPR has also been used in many other areas as a nondestructive investigation tool, such as in geophysics to estimate the structure of Earth sediments and to find the depth of bedrocks and water tables [9]. In archeology, GPR has been used to locate buried structures before digging to prevent accidental damage [10–12]. Moreover, GPR is an effective tool to locate landmines [13–15]. Recently, GPR systems have also found applications in imaging through walls and life detection after disasters [16–20]. 691

692

CHAPTER

15

Ground-Penetrating Radar

As a communication system, a GPR system is a complete system with a transmitter (TX) and a receiver (RX). A normal communication system has an information source, communication media, and a receiver. In wireless voice communication, the source is the voice or data that contains the information to be transmitted by the transmitter and recovered by the receiver. The media of the wireless voice communication system can be anything but is irrelevant to users. The sole purpose of the wireless voice communication system is to correctly send the voice or data to the receiver. In a GPR system, the transmitted source signal is known and therefore contains no information. What is sought are the media characteristics that the GPR waves went through. From the received signal, the GPR system attempts to invert the communication media through which the GPR signals pass. As the media are more complex, GPR systems are also more challenging than voice communication systems.

15.1.1 GPR Operating Principle Also known as material-penetrating radar (MPR), GPR uses electromagnetic radiation in the radio-frequency (RF) and microwave band of the radio spectrum and detects the reflected signals from subsurface structures. It can be used in a variety of media, including rock, soil, ice, freshwater, pavements, and various infrastructures. Detection vectors include anomalies in the soil and changes in material such as layers, voids, and cracks. From a frequency-domain viewpoint, a GPR is an ultrawideband transceiver. A narrow pulse in a time domain contains a wideband frequency spectrum. The narrower the pulse is, the wider the frequency band it occupies. If a transmitter emits a narrow pulse, then it will propagate through the materials to be detected and reflected at the boundaries between different materials with different electrical properties (dielectric constant and conductivities). The time delay between the transmitted and reflected pulses is directly related to the spatial distance between layers as shown in Figure 15-1. The amplitude attenuation, pulse delay, and pulse width expansion are determined by the electrical properties of the materials near the GPR system. Therefore, in the ideal case, narrower pulses are preferred in a GPR transmitter for increased spatial resolution. However, when pulses are narrow, they contain more high-frequency energy and less low-frequency energy due to the limit of the system bandwidth, and therefore the penetration depth will be reduced. So, because the GPR are band limited, appropriate GPR center frequency must be selected in accordance to the application in which it is used. The GPR system uses high-frequency EM waves to probe lossy dielectric materials to detect structures and changes within materials, depending on the waveforms transmitted from a radar transmitter, and can be divided into two categories based on the hardware implementation strategy: impulse [1–4] radar and continuous wave radar. An impulse radar transmitter transmits a high-speed pulse, whereas the continuous wave (CW) radar transmits a continuous wave sweeping frequency in either a linear or stepped mode. In this section, we discuss the impulse radar system, since it is the system used most in GPR. In a pulse radar system, an impulse electromagnetic wave is transmitted from the transmitter antenna and is reflected at the boundaries between two materials with different electrical properties. These waves are received by a GPR receiver and digitized by the control and communication unit, and then finally the data are sent to the host computer for processing and display as shown in Figure 15-1.

15.1

Overview

FIGURE 15-1 ¢ Operating Principle of a GroundPenetrating (or MaterialPenetrating) Radar in a Three-Layer Environment.

Control and Communication Unit

Transmitter (TX)

693

Receiver (RX)

Media I Media II

0.0 in

0.0 ns

5.8 in

2.4 ns

11.5 in

4.8 ns

17.3 in

7.2 ns

23.0 in

9.5 ns

28.8 in

11.9 ns

34.5 in

14.3 ns

40.3 in

16.7 ns

46.0 in

19.1 ns

51.8 in

21.5 ns

57.5 in

23.9 ns

63.3 in

26.2 ns

69.0 in

28.6 ns

74.8 in

31.0 ns

283.67 ft 294.33 ft 305.00 ft 315.67 ft 326.33 ft 337.00 ft 347.67 ft 358.33 ft 369.00 ft 379.67 ft 390.33 ft

As the radar system moves along a designated survey line, an array of waveforms is received by the receiver and displayed in a two-dimensional pseudocolor fashion as shown in Figure 15-2. From Figures 15-1 and 15-2 we can see that the depth range of GPR is limited by the electrical conductivity of the materials, the operating center frequency, and the radiated power of the GPR. As conductivity of the materials increases, the penetration depth decreases. This is because the electromagnetic energy is more quickly dissipated in the conductive materials and converted into heat, causing a loss in signal strength in the media. Higher frequencies do not penetrate as deeply as lower frequencies due to shallower skin depth (which is inversely proportional to the square root of the operating frequency and the conductivity of the materials) but give better spatial resolution. Maximum depth of penetration is achieved in ice and can reach several hundred meters. Good penetration is also achieved in dry sandy soils or massive dry materials such as granite, limestone, and concrete and could be up to 45 meters. In moist or clay-laden

FIGURE 15-2 ¢ Two-Dimensional Pseudocolor Display of Measured Radar Signals.

694

CHAPTER

15

Ground-Penetrating Radar

soils and soils with high electrical conductivity, penetration depth is sometimes only a few centimeters. Ground-penetrating radar antennas are generally in contact with the ground to couple more EM energy to the ground surface; however, air-launched GPR antennas can be used above the ground for shallow detection. Cross-borehole GPR has developed within the field of hydrogeophysics to be a valuable means of assessing the presence and amount of soil water. As shown in Figure 15-1, a GPR system includes four main parts: a transmitter, a receiver, a control unit, and a host computer. The transmitter generates an electrical signal in the frequency band of interest, and the signal is then radiated into material in the form of EM waves by the transmitting antenna. Part of the transmitted wave – the direct wave – reaches the receiving antenna through the air; another part of the wave – the surface reflection wave – is reflected by the ground surface; and the other part of the transmitted wave penetrates the ground surface into the material and is absorbed, transmitted, and reflected back when it hits the interface between the layers. The travel time of each echo is proportional to the distance from the interface to the transmitting and receiving antennas as well as the electrical properties of the media. The receiver receives all these signals. The control unit controls the timing procedures of the transmitter and receiver. The host computer processes the received data and extracts and displays measured pavement information.

15.1.2 Hardware Architectures of GPR Systems To increase spatial resolution, GPR systems have to be ultrawideband. Therefore, for CW radars, some forms of frequency modulation must be applied. Three major architectures have been employed to construct a GPR system: impulse radar, frequencymodulated CW (FMCW) radar, and stepped frequency CW (SFCW) radar. FMCW and SFCW GPRs are frequency-domain implementations of a pulsed GPR. These GPR architectures can be described by the waveform of the transmitted signals. Figure 15-3 shows the transmitted waveforms of different GPR architectures. In the following sections, we will briefly discuss each cited structure.

15.1.2.1 Impulse Radar The transmitted signal of the impulse radar is a nanosecond monopolar or bipolar pulse [Figure 15-3(a)]. The reflected signal is usually a wavelet with a short duration. The delay time of the reflected signal with respect to the direct wave is proportional to the distance from the antennas to the reflection boundary. The impulse radar is the most used GPR architecture due to its hardware simplicity and low cost. However, it has shortcomings such as ultrahigh sampling rate (picosecond equivalent-sampling period), difficulty in accurate frequency band control, and pulse jittering noise. 15.1.2.2 Frequency-Modulated Continuous Wave Radar The FMCW radar radiates a linearly frequency modulated wave [Figure 15-3(b)]. As the frequency of transmitted and received signals increases with the time, delay in the reflected signals is then translated into frequency differences between the transmitted reference signal and the received signals. When the received signal is mixed with the reference signal, which is synchronized with the transmitting signal, a beat signal or

15.1 Pulse

(a)

Amplitude

Overview

1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

0.1

(b)

0.2

0.3

0.4

0.5 Time (s)

FIGURE 15-3 ¢ The Transmitted Waveforms of Different Radar Methods Used in GPR Hardware Implementation.

0.6

0.7

0.8

0.9

1

FMCW (Frequency-Modulated Continuous Wave)

Amplitude

1 0.5 0 –0.5 –1

0.1

0.2

0.3

(c)

0.4

0.5 Time (s)

0.6

0.7

0.8

0.9

1

0.7

0.8

0.9

1

Synthetic Pulse

Amplitude

1

–1

0.1

0.2

0.3

0.4

0.5 Time (s)

695

0.6

frequency-difference signal will be formed. Since the beat frequency is proportional to the time delay and the amplitude is proportional to the strength of the reflected signal, the spectrum analysis of the received signal will determine the target distance and attenuation.

15.1.2.3 Synthetic Pulse Radar Synthetic pulse radar is also called stepped frequency continuous wave radar. In FMCW radar, the frequency of the transmitting signal is linearly modulated and the beat signal spectrum is obtained as a received signal for analysis. Consider a pulsed-GPR transmitting signal: If the time-domain signal is Fourier transformed into the frequency domain, a wide frequency band can be found. This gives us a hint that the pulsed-GPR system can also be implemented in the frequency domain and then inversely Fourier transformed to the time domain, as long as we have enough frequency-domain samples. Therefore, another way to implement pulsed radar is to transmit a single frequency signal each time and step up at a small delta in frequency until the end of the frequency band as shown in Figure 15-3(c). The receiver receives both amplitude and phase at each frequency step. When the received signal is inversely Fourier transformed into time domain, a signal similar to the impulse radar can be obtained. It is clearly seen that SFCW radar is a frequency-domain implementation of the impulse radar. However, SFCW radar has both benefits of FMCW radar and pulsed

696

CHAPTER

15

Ground-Penetrating Radar

radar. Theoretically, in a SFCW radar system, transmitter amplitude, phase, and frequency can be accurately controlled at each frequency step, so it is possible to compensate the nonideal frequency response of antennas to improve spatial resolution and penetration depth. However, the compromise of SFCW radar is its complicated hardware structure and high power consumption. The other consideration is the transmitting power. Pulsed GPR can generate narrow pulses with large amplitude to increase instant power. Most pulsed GPR can reach 100 W of instant power easily. However, for continuous wave radars, the transmitting power is limited to a few watts.

15.1.3 The Pulsed-GPR System As discussed in the previous sections, the GPR system uses pulse echoes to determine material boundaries and characteristics. In Figure 15-1, the directly coupled wave from the transmitter antenna to the receiver antenna is used as a time and amplitude reference since it is least affected by the change in the materials to be measured. The layer and material information is carried in the reflected signals (echoes). We will now discuss echoes in more detail. Consider Figure 15-4, assuming the materials are lossless and the distance between the transmitter and receiver antennas is negligible, the travel distance along this path can be calculated by tc d ¼ pffiffiffiffi 2 er

(15.1)

where t ¼ signal travel time between the transmitting and receiving antenna via the reflecting interface, c ¼ speed of light in a vacuum, and er ¼ relative dielectric constant of the medium. In the ideal case, the reflected signal is just a replica of the transmitted waveform. However, in real application, the received signal not only has the wave component that travels along the path shown in Figure 15-4 but also includes the direct wave and the surface-reflected waves described in Figure 15-1. Figure 15-5 shows a waveform of the received signal in a 400-MHz GPR system that uses a 2-ns pulse transmitter. A0 in Figure 15-5 is the direct coupling between transmitting and receiving antennas, A1 is the reflection from the ground surface, and A2 and A3 are caused by the boundaries between different materials under the ground. The positions of the peaks and the valleys in the FIGURE 15-4 ¢ A Path of Signal Propagation in a One-Layer Structure.

Tx

Rx V

d

εr

t

Pulsed Ground-Penetrating Radar System Design

15.2

Volts

Surface Echo A1 Direct Wave with Attenuation A0

First Interface Return A2

Dt1

Time

Second Interface Return A3

Dt2

received signal represent the travel time of the reflected waves, which is proportional to the distance from the boundaries to the antennas. We also noticed that the actual waveforms transmitted from transmitter antenna, going through materials, and received by the receiver antenna is a typical Gaussian pulse going through a band-limited system. In addition, the actual directly coupled wave A0 has a smaller amplitude because the GPR system has a nonlinear gain-control mechanism to intentionally suppress the directly coupled signal to enhance the reflected signal from layer surfaces. The detection resolution of the radar is determined by the high-frequency components of the received signal. The higher the system frequency, the easier it is to distinguish two closely spaced reflection events. The transmitted signal is usually generated from an avalanche transistor circuit and shaped to a desired frequency band. A typical transmitter signal is shown in Figure 15-6(a). The spectrum of the transmitted signal is shown in Figure 15-6(b). The spectrum of the received signal is determined by the bandwidth of the transmitted signal, the bandwidth of the transmitting and receiving antenna, and the propagation paths. From Figure 15-6, we can see that the highest frequency of the signal reaches to 1.7 GHz, with attenuation of 30 dB from its peak at zero frequency or DC. To process these signals in computers, we need to convert the signals to digital form. Nyquist’s theorem states that if we sample the complex waveform uniformly at a rate more than twice the highest frequency component sine wave contained within, then the samples obtained are sufficient information to reconstruct the waveform. Therefore, the sampling rate for this signal is at least 3.4 GHz to obtain a good time-domain waveform. Thus, a sampling period of at least 300 ps must be achieved.

15.2 PULSED GROUND-PENETRATING RADAR SYSTEM DESIGN 15.2.1 GPR Design Consideration The design of a GPR system is application specific. For different applications, different GPRs with different parameters must be used. Therefore, GPR manufacturing must be custom made for each case. For example, for a road-survey application, an air-launched

697 FIGURE 15-5 ¢ The Waveform of Received Signals in an Impulse GPR System.

Ground-Penetrating Radar

15

(a)

Voltage (V)

FIGURE 15-6 ¢ The Frequency Spectrum of the Received Signal with 2-ns Transmitter: (a) Transmitting Pulse and (b) Spectrum of the Transmitted Pulse.

CHAPTER

(b)

40 35 30 25 20 15 10 5 0

0.5

1.0

1.5

2.0

2.5 3.0 Time (s)

3.5

4.0

4.5

1.4

1.6

5.0

–30 –40 –50

Power (dB m)

698

–60 –70 –80 –90 –100 –110

0.2

0.4

0.6

1 0.8 Frequency (GHz)

1.2

1.8

GPR is used to measure the thickness of the top layer and base layer. The center frequency of this radar should be high enough, usually more than 1 GHz in order to achieve required spatial resolution. The other issue is that the speed of GPR measurements must be high enough for fast data acquisition. The corresponding receiver and data processing should be designed to cope with the system design. The basic system design parameters of a pulsed-GPR system include center frequency, sampling rate, antennas, time window, and equivalent-sampling interval.

15.2.1.1 Center Frequency Selecting the operating frequency for a GPR survey is the first step in the application process. There is a trade-off between spatial resolution, depth of penetration, and system portability. As a rule, it is better to trade off resolution for penetration depth since it is apparently useless to have a great resolution if the target cannot be detected. The best way to approach the problem is to define a generic target type (i.e., point target, rough planar target, or specula target) and specify a desired spatial resolution S. The initial frequency estimate is then defined by the empirical: 150 f ¼ pffiffiffiffi ðMHzÞ S er where er ¼ relative permittivity (dielectric constant) of the material.

(15.2)

15.2

Pulsed Ground-Penetrating Radar System Design

A simple guide is the following table, which is based on the assumption that the spatial resolution required is about 10–40 percent of the target depth as indicated in Table 15-1. When the material is low loss, 10-percent spatial resolution can be achieved. For high-loss materials, upper boundary should be selected. These are values based on empirical data. Since every problem requires careful thought, the values in Table 15-1 should only be used as a quick guide and not as a replacement for thoughtful survey planning. In lighter materials (low loss) such as snow, ice, sand, and rocks, the penetration depth is deeper and the spatial resolution is higher. For a 100-MHz GPR, when used in coal mines, the penetration depth can reach as deep as 20 meters [21]. We should note that in (15.2) the contribution of conductivity of the materials is not shown. This is because the conductivity of the material is difficult to estimate. Actually, the conductivity plays a more important role in spatial resolution and penetration depth. Highconductivity material absorbs more EM energy and reduces penetration depth and spatial resolution. Equation (15.2) applies only to nominal attenuation cases.

15.2.1.2 Window Size The GPR signal is a time-domain signal. The received signal is a time-domain waveform as shown in Figure 15-5. The number of display points is fixed, depending on the GPR clock speed and number of sampling points in one frame. To show the reflection signal from the deepest reflection layer, the GPR window must be set according to the dielectric constant of the material and the investigation depth. If the window is too wide, then the waveforms may be too narrow and detailed information of the subsurface may be lost. However, if the window is too narrow, then only a portion of the total waveform can be obtained. The way to estimate the time window required by a given application can be estimated by the following equation derived from (15.1): W ¼ 1:3

pffiffiffiffi 2 MaxDepth ¼ 8:7 MaxDepth er Velocity

(15.3)

where MaxDepth ¼ maximum depth desired to investigate and Velocity ¼ wave propagation velocity in the media. Usually, to avoid narrow window settings, the preceding expression increases the estimate time by 30 percent to allow for uncertainties in velocity and depth variations. If no information is available about the electrical properties of the subsurface, see TABLE 15-1 ¢ Investigation Depth, Spatial Resolution and Center Frequency of a GPR System in Soil (g ¼ 4.3) Depth of Investigation (m)

Center Frequency (MHz)

Nominal Spatial Resolution (m)

0.5 1.0 2.0 5.0 10 30 50

1,000 500 200 100 50 25 10

0.07 0.14 0.35 0.70 1.41 2.83 7.07

699

700

CHAPTER

15

Ground-Penetrating Radar

Table 15-2, which provides a list of materials and their dielectric constant and conductivity ranges for initial estimation of the velocities of common geologic materials. Note that the attenuation of air is listed as zero, which does not necessarily mean that the wave will transmit in the air without attenuation. As EM wave propagates in the air as a function of distance, the amplitude will reduce inversely proportional to the distance it travels. In other words, the attenuation due to distance is not considered in Table 15-2.

15.2.1.3 Sampling Interval One important parameter in designing a GPR system is the time interval between sampling points on a recorded waveform. The sampling interval is determined by the highest frequency desired in a GPR system. Based on the Nyquist sampling theorem, the sampling frequency of the GPR receiver should be at least twice the highest frequency components desired in the GPR system; in a practical system, it is usually three or more times higher than this minimum requirement. For most GPR antenna systems, the bandwidth to the center frequency ratio is typically about 1. What this means is that the pulse radiated contains energy from 0.5 times the center frequency to 1.5 times the center frequency. As a result, the maximum frequency is around 1.5 times the nominal center frequency of the antenna. TABLE 15-2

¢

The Parameters of Different Geologic Materials

Material

Typical Relative Dielectric Constant

Conductivity (mS/m)

Velocity (m/ns)

Attenuation (dB/m)

Air Distilled water Freshwater Seawater Dry sand Saturated sand Limestone Shale Silts Clays Granite Dry salt Ice

1 80 80 80 3–5 20–30 4–8 5–15 5–30 5–40 4–6 5–6 3–4

0 0.01 0.5 0.01 0.01 0.1–1.0 0.5–2 1–100 1–100 2–1,000 0.01–1 0.01–1 0.01

0.3 0.033 0.033 0.01 0.15 0.06 0.112 0.09 0.07 0.06 0.13 0.13 0.16

0 0.002 0.1 1,000 0.001 0.03–0.3 0.4–1 1–100 1–100 1–300 0.01–1 0.01–1 0.01

TABLE 15-3

¢

Sampling Intervals versus Operating Frequency

Antenna Center Frequency (MHz)

Maximum Sampling Interval (ns)

Minimum EquivalentSampling Frequency (MHz)

10 20 50 100 200 500 1,000

16.7 8.3 3.3 1.67 0.83 0.33 0.17

60 120 300 600 1,200 3,000 6,000

15.2

Pulsed Ground-Penetrating Radar System Design

Based on the assumption that the maximum frequency is 1.5 times the nominal antenna center frequency, data should be sampled at a rate twice this frequency. For good survey design, it is better to use a safety margin of 2. Therefore, the sampling rate should be approximately six times the center frequency of the antenna being utilized. Hence, the sampling frequency of the GPR receiver should be higher than the lower limit of fs 6f0

(15.4)

where fs ¼ sampling frequency and f0 ¼ center frequency of the GPR antenna. The corresponding sampling interval ts is ts

1 6f0

(15.5)

Table 15-3 summarizes suitable sampling intervals versus operating frequency. In modern GPR design, the sampling interval is software selectable and can be largely beyond the numbers listed in Table 15-3. The equivalent-sampling frequency can reach more than 10 times the GPR center frequency. Table 15-4 shows the requirements of GPR transmitter rise time for various GPR center frequencies.

15.2.2 GPR Hardware Block Diagram As shown in Figure 15-7, a complete pulsed-GPR hardware system includes a transmitter antenna, a receiver antenna, a narrow-pulse generator, a sampling bridge, a GPR receiver, various timing and control circuits, a data-acquisition unit, a data-processing and data-communication system, and a data-storage and data-display device. Each block will be described in detail in the following sections.

15.2.2.1 Antenna Selection Consider Figure 15-5. GPR reflections are replicas of the transmitted signals separated by time intervals. The transmitted signal has a main pulse followed by a short tail. The short tail is caused by insufficient bandwidth of the system. Narrowband usually makes the tail of the transmitted signal and reflected signals rather long, resulting in the mix between waves, and therefore reduces the spatial resolution. In an ideal case, GPR system bandwidth should be as wide as possible. Practically, the GPR system bandwidth is about the same as the center frequency – that is, the relative bandwidth is about 100 percent. Although the GPR bandwidth is determined by many factors such as transmitter pulse width, antenna bandwidth, and sampling interval, antenna bandwidth is the most deterministic parameter that plays the bottle neck in a GPR system. Since the TABLE 15-4

¢

GPR Center Frequency versus Transmitter Pulse Width

GPR center frequency (GHz) Required transmitter pulse rise time (ns) First zero crossing frequency in spectrum (GHz)

0.1 0.2 0.4 0.6 1 1.2 2 2.5 3 2.51 1.25 0.63 0.41 0.25 0.21 0.13 0.10 0.08

3.5 0.07

0.40 0.80 1.60 2.39 3.99 4.78 7.97 9.97 11.96 13.95

701

702

CHAPTER

Ground-Penetrating Radar

15

FIGURE 15-7 ¢ Detailed Impulse GPR Hardware Block Diagram.

Control and Communication Communication

Control Unit

Clock Generator and Control

A/D Converter

A

A Slow Ramp Signal

Fast Ramp Signal

Delay Line

C

D Comparator

B

E Pulse Shaper F

Receiver Subnanosecond Pulse Generator

Signal Conditioning

Strobe Generator G

Pulse Shaping and Matching

H Sampling Bridge Sampling Head I

TX TX Antenna

Strobe Trig Control

Antennas

RX Antenna

J Digital Signal Path

Analog Signal Path

spatial resolution is determined by the characteristics of the GPR system bandwidth, antenna selection becomes critical in GPR development. Wideband requirements usually result in a relatively large antenna size to radiate the energy in the spectrum at the low-frequency end. To achieve small antenna dimensions and acceptable antenna performance, the antenna design must compromise between bandwidth and gain for a given application. An important consideration when choosing equipment for any particular application is to determine correctly the exact trade-off between spatial resolution, size of antenna, scope of signal processing, and ability to penetrate the material.

15.2.2.2 High Sampling Rates Data Acquisition The GPR received signal is a high-frequency RF signal. Unlike frequency-domain receivers, the GPR receivers must recover the entire time-domain waveform without distortion. To sample this high-frequency signal, according to the Nyquist sampling concept, the sampling period should be at most half the period of the highest frequency signal in the record. Take the GPR with center frequency of 400 MHz as an example: The maximum spectrum is about 1.7 GHz (Figure 15-6), and therefore a minimum of a 3.4-GHz analog-to-digital (A/D) converter must be used. In circuit design, a 3.4-GHz A/D converter is difficult to implement with the wide dynamic range and low power consumption that a GPR system requires. A non–real-time sampling method is usually used in a GPR implementation to solve this problem. A sample method called equivalent time sampling is used to sample the highfrequency signal without losing the information in the signal but with a relatively low

15.2

Pulsed Ground-Penetrating Radar System Design

sampling rate. Equivalent time sampling, however, is based on the assumption that the received signal is repetitive in the time interval during the sampling period. Detailed discussions about the sequential-sampling method will be introduced in the following sections.

15.2.2.3 Microprocessor and Microcontroller Timing in the GPR system is critical: All the parts – the transmitter, the sampling rates, and the receiver – must follow the sequential clock. A microcontroller is the best way to do this because it has rich peripheral devices such as a timer, a pulse-width modulation wave generator, and a counter, as well as control features necessary in the GPR system design such as filter selection, gain control, window size control, and transmission-delay control. A modern GPR system is so complex that microprocessors become a necessity in the system and timing control. Powerful microprocessors such as the ARM processor and digital signal processor not only can provide control functions but also have additional peripherals such as an Ethernet port, LCD, USB ports, SDRAM drive, and a Flash controller. A microprocessor is generally used to accomplish system control, data acquisition, and data communication. 15.2.2.4 Digital Circuit and Analog Circuit As previously mentioned, timing and control circuits are critical in a GPR system due to requirements for accurate timing and sampling. Logic-integrated chips such as the inverter, OR gate, AND gate, and XOR gate will be used. When the RF signal is obtained from the sampling head, it is very small in amplitude and noisy. Low-noise amplifiers and filters are used to amplify the signal to a certain level and proper frequency band so that the subsequent A/D converters can reach their maximum dynamic range. 15.2.2.5 Communication and Synchronization There are two kinds of communication in the radar system: One is between the transmit and receive antennas, and the other is the communication among the microcontrollers, microprocessors, and host computers. When the transmitter and receivers are located in the same enclosure, the communication and synchronization between the transmitter and receiver is relatively direct. However, in low-frequency GPR systems (100 MHz or lower), to increase detection range and reduce influence from the direct wave, the transmitter and receiver antennas are usually spatially separated. Timing synchronization between transmitter and receiver antennas is critical in such cases. The communication links inside GPR systems must be reliable and chronologically accurate. Communication methods can be wired communication between the microcontroller and microprocessor or wireless communication between antennas and host computer and antenna box. 15.2.2.6 Signal Processing Signal processing here refers to the hardware signal conditioning and software data processing inside the GPR system in real time, not post-processing. For hardware signal processing, it includes the amplifier and filters such as low- and high-pass filters with

703

704

CHAPTER

15

Ground-Penetrating Radar

proper bandwidth and gain selections. Hardware signal processing also includes timevarying gain control to reduce direct waves and to enhance reflected waves. For post-processing, many methods used in seismic signal processing and general signal processing can be adopted, such as background removal, software gain adjustment, software filter, peak tracking, and common middle point.

15.2.2.7 Post-Processing Software Development When a GPR signal is sent to the data-processing unit, usually, a computer, the signal is further processed. In modern GPR systems, a software interface is developed to simplify usage. The interface and post-processing software is usually developed using a higher level language like Visual C þþ , Visual Basic, MATLAB, or C#. These development environments are increasingly user-friendly to developers.

15.2.3 Examples of GPR Applications GPR has found many applications, including void detection, pavement thickness calculation, utility pipe leakage detection, buried pipe and cable locating, peat profiling, coal mining safety inspection, hazardous-waste mapping, oil under ice detection, through-wall imaging, life detection after natural disasters, and archeological imaging. The following sections elaborate.

15.2.3.1 Rebar Imaging In a construction area, reinforced concrete is widely used. When the concrete is mature, it must be inspected to see whether the rebar was placed correctly as part of quality control. In this case, GPR can be used for rebar imaging to display rebar spacing, depth, and even diameter. A color plot of such a measurement on a state highway is shown in Figure 15-8. 15.2.3.2 Pipe Detection Figure 15-9 is the field-test result in detecting pipe in an urban area using a 400-MHz GPR. The display is in full-color mode, and the sampling mode is DMI (distance measurement instrument) trigger, and therefore the horizontal axis is the distance instead of time. The area marked with a circle is the location of a pipe.

FIGURE 15-8 ¢ GPR Image of Highway Rebars. Vertical Axis is the Depth and Horizontal Axis is the Distance in Meters. Each Parabola is a GPR Image of a Rebar reflection. There are Multiple Reflections and Layer Reflections in the Image.

0.0 m 0.6 m

1.2 m

1.8 m

2.3 m

2.9 m 0.0 m

0.3 m

0.6 m

0.9 m

1.2 m

1.5 m

1.8 m

2.1 m

2.4 m

2.7 m 3.0 m

Pulsed Ground-Penetrating Radar System Design

15.2

705

15.2.3.3 Through-Wall Imaging and Life Detection GPR can also be used to detect moving or static images through walls. By processing the measured data, images of the objects on the other side of the wall can be reconstructed. Figure 15-10 shows a typical application of the application of a through wall imaging system. The front-end hardware of the through-wall imager and the life-detection system are very similar to a GPR system. The difference between the two is the signal processing and information display. The life-detection system uses GPR data and conducts timed Fourier transform to detect heartbeat and breath information while the GPR is immobile. The through-wall imaging system, on the other hand, looks for images of the objects on the other side of the wall. Figure 15-11 shows a life-detection GPR system that was used in real rescue operation after the disastrous Wenchuan earthquake on May 12, 2008, in 0.0 in

0.0 ns

5.8 in

2.4 ns

11.5 in

4.8 ns

17.3 in

7.2 ns

23.0 in

9.5 ns

28.8 in

11.9 ns

34.5 in

14.3 ns

40.3 in

16.7 ns

46.0 in

19.1 ns

51.8 in

21.5 ns

57.5 in

23.9 ns

63.3 in

26.2 ns

69.0 in

28.6 ns 31.0 ns

74.8 in 0.08 ft

FIGURE 15-9 ¢ GPR Image of Pipe Detection. The Horizontal Axis shows the Distance in Inches. Gray Scale represents Signal Amplitudes According to the Range of the DataAcquisition System. In this Example, the Maximums are 10 V.

10.67 ft

21.33 ft

32.00 ft

42.67 ft

53.33 ft

64.00 ft

74.67 ft

96.00 ft

Human

Signal Processing

Wall

Breath and Heartbeat

A2 Radar

85.33 ft

A1

A3 A4 Ground

Computer Detail of Radar TX

TX Antenna

RX

RX Antenna

Controller

106.67 ft

FIGURE 15-10 ¢ A Simplified LifeDetection and Through-Wall Imaging System.

706

CHAPTER

Ground-Penetrating Radar

15

Sichuan Province, China. Figures 15-12 and 15-13 are the processed data from a lifedetection GPR system.

15.2.4 GPR Antennas In most GPR systems, the bandwidth bottleneck is the antennas. In antenna theory, the gain and bandwidth product is a constant for a given antenna. Increasing bandwidth also means sacrificing antenna gain. However, due to high instant power, a pulse-GPR FIGURE 15-11 ¢ A 400-MHz LifeDetection GPR Was Used in Rescue Efforts After the Disastrous May 12, 2008, Earthquake in Wenchuan, Sichuan Province, China. Life Signatures under Earthquake Rubble Can Be Obtained by GPR Data Processing [Courtesy of Biken Co.].

FIGURE 15-12 ¢ Processed LifeDetection Radar Signals [Courtesy of Biken Co.].

Trace: 4644 Start

Stop

Breath Signal Extracted

Distance

116

Gain Adjust Meters 1

Breath Rate 0.25 Hz

Motion Signal Extracted

Gain Adjust 1

DC Adjust 1

15.2

Pulsed Ground-Penetrating Radar System Design

707 FIGURE 15-13 ¢ Location of the Person in the RadarDetection Range [Courtesy of Biken Co.].

system usually has enough system gain to cover the loss of antennas. The pulses generated in the GPR transmitter will be sent to the antenna for transmission. As discussed in the previous sections, ideally the pulse radiated from the antennas should be as narrow as possible to obtain a higher spatial resolution. In the meantime, the energy of the pulse must be large enough to penetrate through lossy materials. Therefore, the antenna used in the GPR system should have enough bandwidth to allow fast pulses to be radiated and received. For this reason, a proper compromise must be made in antenna selection. In most cases, it is much easier and more economical to generate high-amplitude pulses than to make a high-gain antenna. In the GPR antenna design, wideband becomes the design objective, and gain is somewhat compromised. GPR antennas are ultrawideband (UWB). For an ultrawideband impulse antenna, the desire is to effectively radiate the pulse signal without distortion. Therefore, the antennas should have a wide working frequency band and a fixed phase center. Since the signal transmitted and received by the antenna usually has a bandwidth that is more than 20 percent of its center frequency, the GPR is defined as an ultrawideband system. The pulse shape carries information of the materials that the waves go through. The transmitted time-domain signal contains abundant frequency components. Thus, the radiation of a GPR antenna depends on how the characteristics change with the frequency in the required frequency band. When an antenna tranceives a single frequency signal, the excitation signal changes continuously and periodically in the time domain. In the pulsed system, we need to consider the relationship between amplitude and phase at the same time in order to minimize the distortion of the pulse due to the antenna characteristics. Consider the Gaussian impulse excitation signal as an example. If the excitation of the antenna is a Gaussian pulse, due to the bandpass nature of the antennas, the radiation field from the antenna is in the differential form of its excitation signal. When the receiver antenna receives this field, if the load is matched with the impedance of the antenna, the received voltage will be completely applied to the load without reflection in the entire band. The current on the load will be proportional to the incoming field. Thus, the received waveform will keep the shape of the differential Gaussian pulse, which means the signal waveform of the receiving antenna

708

CHAPTER

15

Ground-Penetrating Radar

at the maximum receiving direction is similar to the incident signal waveform. However, the characteristics of the GPR antenna are not ideal in practice. We need to consider signal amplitude and phase when designing a GPR antenna according to the frequency response of the antennas. The other issue is that the frequency characteristics of the materials cannot be assumed to be dispersionless. Frequency dispersion will introduce distortions, so distortions may occur during transmission in a wide band in lossy media.

15.2.4.1 GPR Transmitting Pulse The Gaussian pulse signal is a common model in time-domain analysis, which can be expressed by [22]: n tn ! n 2 d eðt=tÞ2 (15.6) gn ðtÞ ¼ n! dtn where n ¼ 0, 1, 2, . . . is the degree of Gaussian signal and t ¼ time constant of pulse width. Figure 15-14(a) shows the waveform in different order when t is equal to 1 ns. By Fourier transformation, the spectral function of the Gauss signal gn(t) is n ! tn pffiffiffiffiffiffiffi 2 ðjwÞn pt2 eðtw=2Þ2 Gn ðwÞ ¼ (15.7) n! We can obtain the maximum frequency of spectral density of the Gaussian pulse in (15.7): pffiffiffiffiffi pffiffiffiffiffi 2n 2n or fc ¼ (15.8) wc ¼ t 2pt If we know the bandwidth and fc, then we can find the degree of the Gaussian pulse by (15.8). By normalizing the spectral density of the signal in (15.7), we can obtain the power spectral density: 8 2 < eðtw=2Þ =2 n ¼ 0 n UðwÞ ¼ (15.9) : e2 2 w2n eðtw=2Þ2 =2 n > 0 4p f c

Figure 15-14(b) shows the power spectral density when t ¼ 1 ns, which shows that when n ¼ 0, the pulse has a large DC component, and when n ¼ 1, the pulse becomes a bandpass signal. In practice, we usually use n ¼ 0 and n ¼ 1 cases to represent pulse generated from a GPR transmitter electronics and the pulse signal received from a GPR receiver, respectively. We notice that when n ¼ 0, the pulse is a unipolar signal in which the DC component is dominant. If the antenna size is too small to

Pulsed Ground-Penetrating Radar System Design

15.2

709

radiate low-frequency signal, then the low-frequency characteristic of the radiation field will compromised. When n ¼ 1, the signal is a differential Gaussian pulse that is bipolar and becomes a bandpass signal, which describes an output when a signal of n ¼ 0 passes through a band-limited system such as an antenna. The Gaussian pulse is a better description of GPR transmitter signal. However, for mathematic simplicity, we can approximate the Gaussian pulse by a triangular signal without losing generality. Figure 15-14(c) shows a triangular approximation of the Gaussian pulse when n ¼ 0, which is a linear approximation of the Gaussian pulse when n ¼ 0.

15.2.4.2 Bow-Tie Antennas Bow-tie antennas [23] are widely used in GPR systems as transmitting and receiving antennas. A bow-tie antenna is a simplification of a biconical antenna. The wideband characteristic and time-domain characteristic of a biconical antenna is ideal (Figure 15-15) because of its characteristics of frequency independency – it is impractical to use, however, due to its size and cost. The space between the two triangle pieces forms a transmission line. If the ratio between the equivalent inductance L and equivalent capacitance C remains constant from the center to the edge, then the characteristic impedance will not change along the transmission line and wideband antenna characteristics will be present. This is similar to the biconical antenna. Although the UWB characteristic of the bow-tie antenna (Figure 15-16) is not as good as that of the biconical antenna, it has a simple mechanical structure, is easy to manufacture, and is easy to integrate into a system. Due to these reasons, the bow-tie antenna is widely used in ground-penetrating radar systems. Note that the bow-tie antenna is a balanced antenna. If we use a parallel line feed, then balanced matching 0

1 Power Spectrum Density/dB

Amplitude/(V/m)

n=1 n=0 0.5 0 –0.5 –1

0.5

1 1.5 2 Time/ns (a) Different degree Gaussian wave form

n=1 n=0

–10 –20 –30 –40 –50 –60

0.5

1 1.5 2 2.5 3 Frequency/GHz (b) Power spectral density of different orders of Gaussian pulse

Amplitude

1

Time (ns) 1

2

(c) Linear approximation of the Gaussian pulse when n = 0 in Figure (a)

FIGURE 15-14 ¢ Waveform of a Gaussian Pulse and Linearized Triangular Pulse.

710

CHAPTER

15

Ground-Penetrating Radar

FIGURE 15-15 ¢ Biconical Antenna.

FIGURE 15-16 ¢ Bow-Tie Antenna.

circuits will be needed. If we use a coaxial line to feed the antenna, then we need to add a balun to keep a balanced input. For GPR applications, the bow-tie antenna can be categorized into three different types: unloaded, end-loaded, and distributed resistive load.

15.2.4.3 Unloaded Bow-Tie Antenna The bow-tie antenna is shown in Figure 15-16. Due to the limited size of the antenna, there is a discontinuity at the end of the antenna, which will generate a reflection (antenna ringing). The reflected signal from the target will mix with the end reflection of the antenna and make the data interpretation difficult. Also note that the unloaded bowtie antenna can be considered a transformed version of a dipole. In most GPR applications, bow-tie antennas are not used without adding resistive loads to reduce the end reflection and increase bandwidth. 15.2.4.4 End-Loaded Bow-Tie Antenna [24] To make the antenna short and minimize the reflections at the end of the antenna, we can add resistive load at the end of the antenna to absorb energy reaching the ends. When the EM energy reaches the resistive load, instead of radiating the energy or reflecting it back, it is absorbed by the resistive load as if the antenna is infinitely long. To minimize the reflection, the value of the resistive load must match the impedance of the antenna ends in the entire frequency band. Therefore, the load impedance must be equal to the characteristic impedance of the antenna. We should notice that the impedance of the antenna is not a constant throughout the frequency band. On the other hand, the end resistor is a constant. The end loading method thus is of limited use when increasing bandwidth. In most cases, the end loading can improve low-frequency performance significantly if proper matching is achieved. To ensure the load current return loop, we can add metal shielding to form a shielded bow-tie antenna. The end-loaded bow-tie antenna can effectively reduce wave reflection. To reduce reflection from the back of the metal shielding, absorbing foam is usually inserted between the back of the antenna and the metal shield.

Pulsed Ground-Penetrating Radar System Design

15.2

711

The load resistor can also cause energy loss and reduce the radiation efficiency. This method actually is a way to trade gain for bandwidth.

15.2.4.5 Definition of Ultrawideband In a communication system, the bandwidth of a signal is defined as the ratio between 3-dB bandwidth and the center frequency, or the ratio between absolute bandwidth and center frequency: B¼

fh fl Df ¼ ðfh þ fl Þ=2 f0

(15.10)

where fh ¼ upper 3-dB frequency, fl ¼ lower 3-dB frequency, Df ¼ fh fl , absolute 3-dB bandwidth, and l f0 ¼ fh þf 2 center frequency. In (15.10), we see that 0 B 2. When fh ¼ fl, the signal is a single frequency signal and B is 0. When the signal high-frequency bandwidth fh is much greater than the low-frequency limit fl, fh >> fl, the signal bandwidth approaches the upper limit of 2. When B < 1%, the signal is a narrowband signal. When 1% B < 25%, the signal is generally considered to be a wideband signal. When B 25%, the signal is defined as ultrawideband. Most GPR systems have a bandwidth of more than 30 percent, and therefore the GPR system is an ultrawideband system. The spatial resolution is determined by the characteristics of the antenna and signal processing employed. In general, to achieve an acceptable spatial resolution, we need a high-gain antenna. This necessitates the antenna dimensions to be able to accommodate the wavelength of the lowest frequency transmitted. To achieve small antenna dimensions and high gain, it is required to use a high carrier frequency, which generally does not penetrate the ground material with sufficient depth. An important consideration when choosing equipment for any particular application is to determine the exact trade-off between spatial resolution, antenna size, method of signal processing, and ability to penetrate the material. Figure 15-17 is a typical GPR antenna with shielding. The resistivity loading is used for increasing bandwidth trading with reduced antenna gain. L R

FIGURE 15-17 ¢ End-Loaded BowTie Antenna with Metal Shield [25].

712

CHAPTER

Ground-Penetrating Radar

15

15.2.5 Design of a GPR Transmitter 15.2.5.1 Discussions of Transmitter Pulse Width Pulsed-GPR transmitter is a high-speed, high-voltage pulse generator. The objective of a pulsed-GPR transmitter design is to obtain a pulse that suits the GPR system antenna. The Gaussian pulse shown in Figure 15-14 is generally assumed in a GPR system. To simplify the analysis, the transmitter pulse is approximated by a triangular wave as shown in Figure 15-18. The triangular wave of the transmitter is shown in Figure 15-18(a); Figures 15-18(b) and (c) are the first- and second-order derivatives of the transmitted wave, respectively. The Fourier transform of the second-order derivative of the transmitted signal can be easily obtained from Figure 15-18(c): FðV 00 Þ ¼ tVp w2 ejwt sin c2

wt

(15.11)

2

Therefore, the spectrum of the transmitter signal can be obtained: FðV Þ ¼

FðV 00 Þ jwt 2 wt ¼ tV e sin c p w2 2

(15.12)

Figure 15-18(d) is the plot of the amplitude spectrum of the transmitter triangular signal. Note that the spectrum peak is at DC with the amplitude of Vpt. The spectrum magnitude is proportional to the area of the pulse. When pulse width is narrow, the amplitude of the spectrum will be small. We also notice that when the pulse is narrow – for example, t is small – the spectrum tends to be flat. The amplitude decays in a squared sinc function with respect to the frequency. In the meantime, we can see that the first zero crossing of the spectrum is f0 ¼ 1/t, which is inversely proportional to half of the pulse duration t, which is the rising edge of the transmitter pulse. If we define the (a)

(b)

V(t)

V ′(t) Vp × 109

Vp t (ns) 1 1

(d)

(c) V″(t) Vp × 109

t (ns) 1 –2Vp × 109

2

–Vp × 109

2 T (ns)

2

Amplitude

FIGURE 15-18 ¢ GPR Transmitter Pulse and Its Spectrum.

Spectrum of Triangular GPR Pulse × 10–7 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 2.5 3 0 0.5 1 1.5 2 Frequency (Hz) × 109

15.2

Pulsed Ground-Penetrating Radar System Design

maximum GPR system bandwidth as equal to the first zero crossing of the transmitted pulse spectrum, then the maximum GPR bandwidth can be written as Bmax ¼

1 t

(15.13)

We must note that the spectrum of an impulse generated by a GPR transmitter has its major energy at DC. As previously discussed, the pulsed GPR directly transmits the impulse from the transmitter antenna without modulation. The portion of the radiated signal is only a part of the entire transmitted spectrum. When the pulse width is narrow, the spectrum tends to be flat, which means that the spectrum includes more highfrequency components. The antenna bandwidth is usually much narrower than that of the transmitted pulse. However, when the pulse rise time is very slow, the highfrequency components will have small amplitude. Due to hardware limitations, it is difficult to create narrow pulses with high amplitude. In designing a GPR transmitter, we must optimize the output energy at the specified frequency band by correctly compromising between the amplitude of the pulse and the pulse width of the transmitter pulse, which are determined by transmitter hardware. The 3-dB bandwidth of the pulse shown in Figure 15-18 can be calculated by solving wt 1 ¼ pffiffiffi (15.14) sin c2 2 2 The solution can be easily found to be Bpulse ¼

1:8955 pt

(15.15)

When the pulse amplitude reduces to half its maximum, then we define the frequency as half the amplitude frequency, f1/2: f1 ¼ 2

1:392 pt

(15.16)

where Bpulse is the 3-dB bandwidth of the pulse shown in Figure 15-18. If we define the GPR center frequency equal to Bpulse, then we can calculate the required pulse rise time of the GPR transmitter. This definition will result in a very fast transmitter pulse, which will reduce the pulse amplitude significantly. In practice, a reasonable compromise is to select a GPR frequency equal to the frequency point that the pulse amplitude reduces to half its maximum. Based on this definition, Table 15-5 lists the pulse rise time required for typical GPR center frequencies. From Table 15-5, we can see that in transmitter circuit design, the rise time of the transmitted pulse determines the frequency band of a GPR system. Something else to note is that as long as the first zero-crossing frequency is within the GPR antenna band, the GPR can still radiate energy to the media with compromised radiation energy. The steeper the slope and the shorter the pulse duration, the more high-frequency components the system will have. These high-frequency components determine the system’s spatial resolution. The amplitude of the transmitted pulse, on the other hand, affects the signal-to-noise ratio (SNR) of the received signal: The higher the amplitude, the greater the SNR of the receiver. However, increased transmitter amplitude will

713

714

CHAPTER

15

Ground-Penetrating Radar

introduce more noise into the system through the trigger paths and will raise the cost of the transmitter circuit and reduce the maximum pulse-repetition frequency (PRF). Sampling circuits in most pulse-GPR systems are based on a sequential-sampling technique. The duration of the sampling pulse applied to the sampling circuit holds the key to the time resolution of the received signal. In receiver circuit design, the systems with shorter aperture time will have a higher time resolution, which also implies the better spatial resolution. The amplitude of the sampling pulse defines the dynamic input range of the sampling circuit. Subsequently, it affects the SNR of the GPR system. Sampling systems with higher control pulse amplitudes will provide a wider dynamic input range. Unfortunately, shorter pulse duration and higher pulse amplitude are contradictory parameters in circuit design. PRF is another concern Trade-offs among all these parameters have to be made in the system design.

15.2.5.2 Methods of Generating Nanosecond Pulse The amplitude of the variable nanosecond pulse mainly affects the system in two ways. In a GPR transmitter, the amplitude of the transmitted pulse affects the SNR of the received signal. The higher the amplitude, the better the SNR of the receiver will reach. The PRF is another concern. It affects the resolution of the receiving signals. Trade-offs among all of these parameters are another necessity. Three different methods are used to generate nanosecond pulses: (1) avalanche transistor pulse generator, (2) emitter-coupled logic pulse generator, and (3) pulse generator using high-speed gates [26, 27]. A brief comparison of these three methods is shown in Table 15-5. In the next section, we will discuss the method of using an avalanche transistor to implement the differential pulse generator as the transmitter due to its simplicity and effectiveness. 15.2.5.3 Avalanche Transistor Pulse Generator Transistors biased in the ‘‘avalanche region’’ may show a negative differential resistance between the collector and the emitter. This part of a bipolar transistor is used for nanosecond pulse generation. The avalanche transistor circuit is a traditional method used to generate a nanosecond pulse. As shown in Figure 15-19, the normal range of transistor TABLE 15-5

¢

The Comparison of the Three Nanosecond Pulse-Generating Methods Avalanche Transistor Pulse Generator

Emitter Coupled Logic Pulse Generator

High-Speed Logic Pulse Generator

Amplitude of the pulse Rising time of the pulse Pulse repetition rate Jitter Output format

High > –20 mA –I(C2)

IC(Q2)

10 V

5V

0V 20 ns V(V4:+)

V(V3:+)

40 ns V(V3:+) V(Q2:C) Time

60 ns

70 ns

722

CHAPTER

Ground-Penetrating Radar

15

In Figure 15-26, Q2 is a high-speed bipolar transistor with a short turn-off time and small output capacitance. The operational amplifier (op amp) AD818 can provide enough linear working range. From Figure 15-27, we can see that the shutdown transition time of the transistor is only about 2 ns from point A to point B. In the transition period, nonlinearity will be introduced only in a very short time range at the beginning. After point B, the current goes to the switch, which is represented by the grey curve in the upper plot in Figure 15-27, is almost zero, and all current goes to the charging capacitor, which is represented by the black curve in the upper plot. Together, these improve the linearity from point B to point C.

15.2.7.2 Design of Slow Ramp Signal Generator The slow ramp signal is a stair-type signal that is used to control the value of increment delay between every two pulses in the sample pulse train. A digital-to-analog converter (DAC) can be used for this task. For example, an AD7533 can serve to generate the slow ramp signal because it is easy to control the increment. The schematic of the slow ramp generator is shown in Figure 15-28. The reference voltage is sent to AD7533 from port E and op amp U33A is used as an inverting amplifier to turn the positive reference voltage to a negative one. The output voltage of the inverting amplifier is calculated by VH ¼

R47 VE R52

(15.19)

The equivalent circuit of the DAC, AD7533, is shown in Figure 15-29. The output voltage can be calculated by Vin ¼ VH

N 1024

(15.20)

Then the circuit can be simplified and analyzed by using the schematic shown in Figure 15-30. +12V_ A

8

2

+

OUT

H 1

U33A LF353

R52

E

8

C175

–12V A C149

+

4

7

15

R47

V– V+

1

C135

C130

5

U34B LF353 6

V– OUT V+

ad7533

100 mv

G

16

3

D10(L) Vdd GND D9 D8 Rfb D7 D6 lo1 D5 D4 lo2 D3 D2 D1(M) Vref

C131

C 165

P B0 P B1 P A0 P A1 P A2 P A3 P A4 P A5 P A6 P A7

14 3

R32

R33

–12V_A4

U11 13 12 11 10 9 8 7 6 5 4

+12V A C 164

2

FIGURE 15-28 ¢ Schematic of the Slow Ramp Generator.

C139

F

Pulsed Ground-Penetrating Radar System Design

15.2

Rfb

723 FIGURE 15-29 ¢ The Equivalent Circuit of AD7533.

R29 R R30

R

Vin

Iout1

Iout2

Iout2

R14 1k

+12 V

+12 V

FIGURE 15-30 ¢ The Equivalent Circuit of the Slow Ramp Signal Generator.

5

+ V+

100 mV

8

R15 100 k

U11B R17

R

LF353 6

R

C45 2n

–12V

R16

Vout

– 4

Vin

7

V–

OUT

From the equivalent circuit, the voltage at point F in Figure 15-28 can be calculated by VF VG VG VIN ¼ R29 R30

(15.21)

Since R29 ¼ R30 ¼ 10 kW, we have VF ¼ 2 VG VIN ¼ 2 VG VH

N R47 N ¼ 2 VG þ VE (15.22) R52 1;024 1;024

From (15.22) we can see that VG sets up the offset voltage of the slow ramp signal. The feedback capacitor C45 in Figure 15-30 forms a low-pass filter (LPF) along with the op amp, which greatly reduces the power-supply noise coupled from E in Figure 15-28. The measured waveform of the slow ramp signal is shown in Figure 15-31. When feeding the fast ramp signal and the slow ramp signal input into the comparator, a pulse train with a controllable increment between each two adjacent pulses is generated. This pulse train will be used for the sampling head as the sequential-sampling trigger command.

724

CHAPTER

15

Ground-Penetrating Radar

FIGURE 15-31 ¢ The Measured Waveform of the Slow Ramp Signal.

15.2.7.3 Design of Delay Line As shown in Figure 15-24, the transmitter pulse and receiver sampling command must be synchronized so that the sampling bridge is triggered at the time when the received signal arrives at the receiver front end. Note that the received signal exists only for several tens of nanoseconds, determined by the window size, and the transmitter and receiver delays are quite different. The receiver circuits have longer delay time than the transmitter circuits due to the complexity of the receiver system. To balance the delay difference between transmitter system and receiver circuits, an additional delay circuit is inserted into the transmitter path. The main purpose of the delay line circuit in Figure 15-24 is to adjust the sampled waveform in the sampling window. From Figure 15-24 we can see that the sampling pulse train to the sampling head will have a certain time delay as compared with the trigger at point B. This time delay is controlled by the R, C, and chips in this signal path and is a fixed value for each system, which is around 50 ns to several 100 ns. If no delay line is used, then the reflecting waveform may be out of the sampling window. As such, the delay line is needed to make sure the whole reflecting waveform can be sampled at the sampling head. Figure 15-32 shows the simulation model of the delay line circuit, and the simulation result is shown in Figure 15-33. If the C1 and C3 are fixed, then the output delay is only controlled by the resistors R1 and R3. The simulation model has two delay channels to compare the delay differences of different circuit parameters. 15.2.7.4 Design of Sample and Hold Circuit As previously discussed, the output waveform of the sampling head is a series of spikes, of which the envelope is the ‘‘down-converted’’ high-frequency waveform. In this case, a sample-and-hold circuit is needed to hold the maximum value of each spike. The sample-and-hold circuit is the basis of the data-acquisition process. Basically, an S/H amplifier circuit has two basic and distinct operational states: sample and hold. In the sample state, an input signal is sampled and simultaneously transmitted to the output. While in the hold state, the last value of the input signal is held until the sample command is reissued. The purpose of the S/H is to maintain the analog input voltage at a constant level for the period of time required to perform an A/D conversion. In the GPR system, the S/H ‘‘freezes’’ the peak voltage of the sampled pulses at the time of the hold

15.2 U1C 5

6 74HC14

Pulsed Ground-Penetrating Radar System Design R3 3 3k C3 100 p

U2B C4 4 74HC14 100 n

R4V 1k

725 FIGURE 15-32 ¢ Simulation Model of the Transmitter Delay Line Circuit in a GPR Receiver.

0 V1 = 0 V2 = 3.3 V TD = 100 ns TR = 200 ns TF = 200 ns PW = 2400 ns PER = 5000 ns

1 V1

V

U1A 2 74HC14

R1 3 1k C1 100 p

U1B 4 74HC14

C2

100 n

R2V 1k

Bias (active) 4.975 v

4.000 v

3.000 v

2.000 v

1.000 v

0v

–1.000 v 0.000 us 0.500 us 1.000 us 1.500 us 2.000 us 2.500 us 3.000 us 3.500 us 4.000 us Time V(R2:2) V(U1A:2) V(R4:2)

command and keeps this voltage unaltered until the next sampled command comes. By doing this, a time-stretched GPR waveform is generated. Figure 15-34 shows the input and output waveforms of the sample-and-hold circuit. The input signal is in blue and the output in yellow. There are many noise sources in this circuit. The main noise source is the sampling pulses. Acquisition time is the time required for the S/H to acquire and then track the input signal after the sample command. Acquisition should end when the signal has settled and remain within the rated error band. This is critical in the system adjustment. By adjusting the time constant of the RC circuit, the delay between sampled signal pulses and the S/H trigger can be adjusted and optimized.

15.2.7.5 Design of the Low-Pass Filter and Amplifier The output waveform of the sample-and-hold circuit is noisy due to high-frequency sampling. A low-pass filter is needed to eliminate the high-frequency noises. Subsequently, an amplifier is needed to adjust the amplitude to increase the dynamic range.

FIGURE 15-33 ¢ Simulation Results of the Delay Line Circuit.

726

CHAPTER

15

Ground-Penetrating Radar

FIGURE 15-34 ¢ The Input and Output Waveforms of an S/H Circuit in a GPR Receiver.

10

FIGURE 15-35 ¢ Input Waveform of LPF.

8 6 4

Volt

2 0 –2 –4 –6 –8 –10

1,000

2,000

3,000

4,000

5,000

6,000

7,000

8,000

9,000

10,000

Number of Sampling Point

There are several types of LPFs, including the Butterworth filter, the Chebyshev filter, the Bessel filter, and the Gaussian filter; each has a different performance and application. In terms of GPRs, we do not want the LPF to introduce distortion into the waveforms, so a safe choice is the Butterworth filter, of which the magnitude of the in-band gain is maximally flat. The input waveform is shown in Figure 15-35. MATLAB is used to optimize the bandwidth of the LPF, which is shown in Figures 15-36–15-38.

15.2

Pulsed Ground-Penetrating Radar System Design

10

727 FIGURE 15-36 ¢ MATLAB‡ Simulation Results with Second-Order LPF; the Bandwidth Is 20 kHz.

8 6

Volt

4 2 0 –2 –4 –6

100

200

300

400

500

600

700

800

900

1,000

Number of Sampling Point

8

FIGURE 15-37 ¢ MATLAB‡ Simulation Results with Second-Order LPF; the Bandwidth Is 10 kHz.

6 4 2 0 –2 –4 –6 –8

100

200

300

400

500

600

700

800

900

1,000

After comparing the simulation results with LPF with different bandwidths, the 10-kHz bandwidth is selected because it gives a smooth waveform without losing many details on each waveform signature. A more systematic way to design the filter bandwidth is to Fourier transform the GPR waveform in Figure 15-38 and design the filter based on the spectrum distribution of noises. Figure 15-39 shows the modified Sallen–Key LPF circuit. The circuit is named after the authors of a 1950s paper describing the technique. Compared with the passive filter with the same order, it provides a higher Q (Quality Factor) by using a positive feedback amplifier with a simple structure. If the positive feedback is properly designed, almost

728 FIGURE 15-38 ¢ MATLAB‡ Simulation Results with Second-Order LPF; the Bandwidth Is 5 kHz.

CHAPTER

Ground-Penetrating Radar

15

5 4 3 2 1 0 –1 –2 –3 0

100

200

FIGURE 15-39 ¢ Modified Sallen–Key Circuit.

300

400

500

600

900

1,000

R7

8

R6

800

C14

Vf

Vi

700

Vp

3

+

C20

OUT

2

1

Vo

– 4

Vn

U6A V– V+

–4

R4 R3

any Q can be realized, limited mainly by the physical constrains of the power supply and the component tolerance. The circuit shown is a two-pole low-pass filter that can be configured as any of the three basic types: Chebyshev, Butterworth, or Bessel. The mathematical equation can be driven as follows. As shown in Figure 15-39, by using Kirchoff’s current law at Vf, Vp, and Vn, we get the following three equations:

1 1 1 1 þ Vp þ Vo ðsC14 Þ þ þ sC14 ¼ Vi Vf R6 R7 R6 R7 1 1 Vp ) Vf ¼ Vp ð1 þ sR7 C20 Þ þ sC20 ¼ Vf R7 R7 1 1 1 R3 ¼ Vo ) Vn ¼ Vo þ Vn R3 R4 R4 R3 þ R4

(15.23) (15.24) (15.25)

15.2

Pulsed Ground-Penetrating Radar System Design

Then Vp can be solved by using (15.23) and (15.24) together: 0 1 R7 B C s2 C14 C20 C Vp ¼ Vi B 2 @ R A R6 R7 R6 R7 R7 7 2 þ þ þ þ R R 6 7 s2 C14 C20 sC14 sC20 sC14 0

1 R6 R7 B C sC20 C þ Vo B 2 @ R A R6 R7 R6 R7 R7 7 2 þ þ þ þ R R 6 7 s2 C14 C20 sC14 sC20 sC14

(15.26)

Then the ideal transfer function can be driven by setting Vp ¼ Vn. By doing this, we get H(s): HðsÞ ¼

Vo K ¼ Vi 1 þ sðC20 R6 þ C20 R7 þ C14 R6 ð1 KÞ þ s2 C14 C20 R6 R7

(15.27)

The standard frequency-domain equation for a second-order LPF is HðsÞ ¼

K

2 jf f 1þ Qfc fc

(15.28)

where fc ¼ corner frequency and Q ¼ quality factor. Following (15.28), we can set 1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2p R6 R7 C14 C20 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi R6 R7 C14 C20 Q¼ C20 ðR6 þ R7 Þ þ R6 C14 ð1 K Þ fc ¼

(15.29) (15.30)

By setting R6 ¼ R7 ¼ R, C20 ¼ C14 ¼ C, the fc and Q can be written as follows: fc ¼

1 2pRC

(15.31)

1 3K

(15.31)

There is another way to calculate the parameters. We still set R6 ¼ R7 ¼ R, C20 ¼ C14 ¼ C. Then R ¼ k1/C*Fcutoff, and R4 ¼ R3k2. Table 15-6 provides the necessary information to select the appropriate resistors and capacitors. Figure 15-40 shows the input and output waveforms of the second-order Sallen–Key low-pass filter with 10-kHz bandwidth. We can see that the LPF not only removes the

729

730

CHAPTER TABLE 15-6

Ground-Penetrating Radar

15 ¢

Parameters for designing Bessel, Butterworth, and Chebychev filters Bessel

Butterworth

Chebychev

# Poles

k1

k2

k1

k2

k1

k2

2 stage 1 4 stage 1 stage 2 6 stage 1 stage 2 stage 3 8 stage 1 stage 2 stage 3 stage 4

0.1251 0.1111 0.0991 0.0990 0.0941 0.0834 0.0894 0.0867 0.0814 0.0726

0.268 0.084 0.759 0.040 0.364 1.023 0.024 0.213 0.593 1.184

0.1592 0.1592 0.1592 0.1592 0.1592 0.1592 0.1592 0.1592 0.1592 0.1592

0.586 0.152 1.235 0.068 0.586 1.483 0.038 0.337 0.889 1.610

0.1293 0.2666 0.1544 0.4019 0.2072 0.1574 0.5359 0.2657 0.1848 0.1582

0.842 0.582 1.660 0.537 1.448 1.846 0.522 1.379 1.711 1.913

FIGURE 15-40 ¢ The Input and Output Waveforms of the Second-Order Sallen–Key Butterworth LPF with Bandwidth ¼ 10 kHz.

high-frequency component generated in the S/H procedure but also removes part of the noise from the previous circuit.

15.2.7.6 Design of Time-Varying Gain Amplifier As shown in Figure 15-5, the waveform of the received signal of the GPR system can be divided into three parts according to the sources of the signal. The first part in the signal waveform, which comes from the direct coupling between the transmitting and receiving antennas at a time range from 0 to 5 ns, carries no information from the subsurface structure but has a very large magnitude. The second part of the waveform, which is reflected by the ground surface, indicates the position of the ground surface and has a relatively large magnitude. The third part in the waveform, which comes from reflections on the boundaries of the underground structures, carries the useful information about the subsurface structure but has a low amplitude.

15.3

GPR System Implementation and Test Results

731

From Figure 15-5 we can see the amplitude of each waveform. These waveforms represent the reflections from subsurface features, and they are separated in time. The closer target has greater amplitude, and the reflection waveform comes near the direct wave. For the amplifier circuit, if a higher gain is applied to the part of the received signal that carries useful information of the underground structure, such as the third part in the received signal, then a larger dynamic range, deeper detection range, and higher SNR can be obtained, which means that the detecting range can be increased when we keep the same spatial resolution. A time-varying gain amplifier is an effective way to implement this idea. As an example, the GPR system can use LMH6503 as the variable gain-control amplifier, and an Microprocessor Unit (MCU) is used to generate the gaincontrol signal.

15.3 GPR SYSTEM IMPLEMENTATION AND TEST RESULTS 15.3.1 GPR System Implementation GPR hardware can be implemented with different center frequencies. Typical commercial GPR systems have center frequencies of 50 MHz, 100 MHz, 200 MHz, 400 MHz, and 1 GHz. As discussed in previous sections, higher GPR center frequencies result in better spatial resolutions with compromises in penetration depth. Figure 15-41 to Figure 15-43 show the measured results at the same site using 200-MHz, 400-MHz,

FIGURE 15-41 ¢ Measured GPR Data over a Section of Driveway Using a 200-MHz GPR System.

FIGURE 15-42 ¢ Measured GPR Data over a Section of Driveway Using a 400-MHz GPR System.

732

CHAPTER

15

Ground-Penetrating Radar

FIGURE 15-43 ¢ Measured GPR Data over a Section of Driveway Using a 1-GHz GPR System.

FIGURE 15-44 ¢ The Internal Structure of a 200-MHz GroundCoupled GPR System with Shielded Bow-Tie Antennas.

Receiver Control and Communication

Transmitter Battery Power

Shielded Antenna Box

and 1-GHz GPR systems. Clearly, the 1-GHz GPR system has the highest spatial resolution while the 200-MHz GPR has the deepest penetration. A ground-coupled 200-MHz GPR hardware internal structure with shielded bow-tie antennas is shown in Figures 15-44. Note that all the circuits are shielded to avoid interference from high-frequency signals. Figure 15-45 and Figure 15-46 show the structure and assembled 400 MHz GPR and 1 GHz GPR units, respectively. The air-coupled GPR is used for top-surface thickness detection, and the required depth resolution is less than half an inch. Figure 15-47 shows an example of an aircoupled GPR unit mounted on a vehicle for road survey. The specifications for the 200-MHz, 400-MHz, and 1-GHz GPR systems are listed in Table 15-7. The transmitter pulse waveforms of different transmitters are shown in Figures 15-48, 15-49, and 15-50, respectively. The waveform of differential strobe pulses for sequential-sampling heads is shown in Figure 15-51.

GPR System Implementation and Test Results

15.3

733 FIGURE 15-45 ¢ The Internal Structure of a 400-MHz GroundCoupled GPR System with Shielded Bow-Tie Antennas.

FIGURE 15-46 ¢ The External Connections of a 1-GHz GroundCoupled GPR System with Shielded Bow-Tie Antennas.

Wired Connection

On–Off Switch

Battery Charger

Handle

Enclosure Box

FIGURE 15-47 ¢ The 1-GHz AirCoupled GPR System Mounted on a Vehicle.

Cable

Mounting Frame

1-GHz AirCoupled GPR

734

CHAPTER

15

Ground-Penetrating Radar

TABLE 15-7 ¢ The Specifications of the Different Sub-Nanosecond Differential Pulse Generators for Sampling Head and Different Transmitters Sampling head

200-MHz transmitter

400-MHz transmitter

1-GHz transmitter

FIGURE 15-48 ¢ The Output Waveforms of the Differential 1-ns Transmitter Used in the 1-GHz GPR Transmitter Measured by a Tektronix 11801B Sampling Oscilloscope with Attenuator.

FIGURE 15-49 ¢ The Output Waveforms of the Differential 1-ns Transmitter Used in the 400-MHz GPR Transmitter Measured by a Tektronix 11801B Sampling Oscilloscope with Attenuator.

PRF Sample points per cycle Strobe pulse rising time Strobe pulse rising time Strobe pulse amplitude Pulse rising time Pulse falling time Pulse width Pulse amplitude Pulse rising time Pulse falling time Pulse width Pulse amplitude Pulse rising time Pulse falling time Pulse width Pulse amplitude

200 kHz 1,024 371 ps 334 ps 7 V 2.3 ns 3 ns 5.3 ns 150 V 860 ps 960 ps 2 ns 58 V 430 ps 514 ps 1 ns 30 V

15.3

GPR System Implementation and Test Results

735 FIGURE 15-50 ¢ The Output Waveforms of the Differential 1-ns Transmitter Used in the 1-GHz GPR Transmitter Measured by a Tektronix 11801B Sampling Oscilloscope Attenuator.

FIGURE 15-51 ¢ The Waveforms of the Differential Strobe Pulses for Triggering the SequentialSampling Gates in a 400-MHz GPR Receiver.

The waveforms recovered by the sampling head of these 400-MHz and 1-GHz pulsed radar systems are shown in Figures 15-52 and 15-53. For each of these figures, both waveforms recovered by the sequential-sampling head we designed and by the oscilloscope TDS3052B are shown. Those figures prove that the variable nanosecond differential pulses generator works well for the pulsed-GPR system up to 1 GHz. They also prove that these GPR systems work well and that the matching for the transmitter to the transmitting antenna and for the receiving antenna to the sampling head is relatively accurate for the ultra wideband. There are however, some differences between the two recovered waveforms.

15.3.2 GPR System Tests After system implementation, tuning, and optimization, lab tests should be done to test the performance of the GPR systems.

15.3.2.1 Air Tests The air tests are designed to test the GPR’s maximum detection range and linearity. A metal plate can be set in front of the GPR being tested and then moved away from the GPR in an open area. One test site in a crowded urban area is the roof of a building. This will allow the detecting ranges in the air for different GPR systems to be measured. Figures 15-54 to 15-56 show the test results. The reflection lines have wiggles, indicating that the metal plate stopped or moved more slowly.

736

CHAPTER

15

Ground-Penetrating Radar

FIGURE 15-52 ¢ Recovered Waveforms of the 400-MHz PulsedGPR System.

FIGURE 15-53 ¢ Recovered Waveforms of the 1-GHz Pulsed-GPR System.

15.3.2.2 Field-Test Comparison of GPR System with Different Frequencies GPR systems with different center frequencies show the capability of different detecting depths and spatial resolutions. In this section, we compare the differences by using the GPR systems of different center frequencies at the same test conditions. Two locations are used for the comparisons of GPR systems with three different working frequencies: 400 MHz, 600 MHz, and 1 GHz. Figures 15-57 through 15-59 show the measured raw data in location 1. Figures 15-60 through 15-62 show the measured raw data in location 2. From the six raw data measurements taken in the two locations, we can see that the 1-GHz GPR provides the best resolution while the 400-MHz GPR has the deepest detection. Table 15-8 lists the measured depths and resolutions for the different GPR systems. TABLE 15-8 ¢ Measured Depth Detecting and Spatial Resolution for Different GPR Systems GPR System

Depth Detecting

Spatial Resolution

200-MHz GPR 400-MHz GPR 600-MHz GPR 1-GHz GPR

Principles of Modern Radar - Volume 3 - PDFCOFFEE.COM (2024)
Top Articles
Latest Posts
Recommended Articles
Article information

Author: Laurine Ryan

Last Updated:

Views: 5804

Rating: 4.7 / 5 (57 voted)

Reviews: 88% of readers found this page helpful

Author information

Name: Laurine Ryan

Birthday: 1994-12-23

Address: Suite 751 871 Lissette Throughway, West Kittie, NH 41603

Phone: +2366831109631

Job: Sales Producer

Hobby: Creative writing, Motor sports, Do it yourself, Skateboarding, Coffee roasting, Calligraphy, Stand-up comedy

Introduction: My name is Laurine Ryan, I am a adorable, fair, graceful, spotless, gorgeous, homely, cooperative person who loves writing and wants to share my knowledge and understanding with you.