Design and Analysis of Modern Tracking Systems Samuel Blackman Robert Popoli IH Artech House Boston @ LondonContents =a ae PW Hwy a 52 14.3 14.4 145 146 15 16 16.1 162 Preface xxvii Acknowledgments yout The Basics of Target Tracking 1 Introduction 1 Basic Processing Definitions 4 Elements of a Conventional MTT System 5 Measurement Data Processing 6 Data Association 8 Filtering and Prediction il Overview of Data Association Issues and Methodologies 12 Interpretation and Issues 12 Unique-Neighbor Versus All-Neighbors Data Association 15 Sequential Versus Deferred Decision Logic 16 Incorporation of Group Information . 7 Use of Thresholding 18 State Estimation Without Explicit Data Association 19 Multiple Sensor Considerations 19 System Design Issues 21 Joint Sensor/Tracking System Design 22 Performance Evaluation 23 viiDesign and Analysis of Modern Tracking Systems 1.7 Related Background Reading 24 References 24 2 Sensor and Source Characteristics 29 2.1 Preliminaries 32 2.1.1 Kinematic Measurements 32 2.1.2 Attribute Measurements 33 2.1.3 Signal-to-Interference Ratio and Measurement Availability and Quality 35 2.2 Tracking Radar Sensor Systems 36 2.2.1 Radar Sensor Overview 36 2.2.2 Measurement Availability 39 2.2.3 Measurement Quality 68 2.2.4 False Measurements and Other Measurement Disturbances 73 2.2.5 Further Measurement Signal Characteristics 76 2.2.6 Sensor Agility 80 92.7 Sensor Management Information 81 2.3 Infrared Search and Track Systems 85 2.3.1 Electro-Optical Sensor Overview 85 23.2 IRST Sensor Overview 86 2.3.3 The Nature of IR Radiation 89 23.4 Target Energy 94 23.5 Energy Propagation and Atmospheric Effects 100 23.6 Background Energy 104 2.3.7 Measurement Availability 107 9.3.8 Increasing Measurement Availabilicy With Signal Processing 112 2.3.9 Kinematic Measurement Accuracy 115 2.3.10 Attribute Measurement Capabilities 115 23.11 Sensor Artifacts 117 2.3.12 IR Processing Chain 117 2.4 Electronic Support Measures Sensor Systems 119 2.4.1 ESM Sensor Overview 119 2.4.2 ESM Sensor Signal Processing 121 2.4.3 ESM Measurement Availability 122 2.44 ESM Measurement Utilization and Interpretation 123 2.5 Laser Sensor Tracking Systems 124 2.5.1 Laser Sensor Overview 124Contents x 2.5.2 — Laser Sensor Measurements 124 2.5.3 Laser Sensor System Applications 127 2.6 — Bistatic Radar Tracking Systems 128 2.6.1 Bistatic Radar Overview 128 2.6.2 Bistatic Radar Measurements 130 2.6.3 Bistatic Radar Tracking 131 2.7 Acoustic Sensor Tracking Systems 131 2.4.1 Acoustic Sensor Overview 131 2.7.2 Acoustic Sensor Measurement 133 2.8 Warning Sensor Systems 134 References 136 Appendix 2A: Radar Pp and Pra Calculations 141 3 Kinematic State Estimation: Filtering and Prediction 147 3.1 Introduction 147 3.2 Least Squares Estimation 148 3.2.1 Linear Least Squares 148 3.2.2 Nonlinear Least Squares Estimation 151 3.2.3 Recursive Least Squares 155 3.2.4 Observabiliry 157 3.3 Kalman Filtering 157 3.3.1 Kalman Filter Definition 158 3.3.2 Correspondence Between Kalman and Fixed-Gain Filters 160 3.3.3 Example of a Two-State Kalman Filter 161 3.3.4 System Driving Noise and Maneuver Model 163 3.4 Extended Kalman Filtering 164 3.5 Example Comparing Nonlinear Estimation Performance 168 3.5.1 Problem and Method Definitions 168 3.5.2 Results 170 3.6 Nonlinear Filtering With Pseudomeasurements 172 3.7 Choice of Tracking Coordinate System 174 3.7.1 Choice of Origin and Axes 174 3.7.2 Cartesian Coordinate Tracking 178 3.7.3 Spherical/Sensor Coordinate ‘Tracking 182 3.7.4 Cartesian Angle-Only Filtering 187Design and Analysis of Modern Tra cking Systems 3.8 3.8.1 3.8.2 3.9 41 42 421 42.2 4.2.3 4.24 43 43.1 4.3.2 4.3.3 44 44] 442 4.43 444 45 4.5.1 452 4.53 4.5.4 45.5 4.5.6 4.6.1 4.6.2 47 48 48.1 4.8.2 Filter Simplification 189 State Reduction 189 Filter Decoupling 191 Conclusions and Future Directions 193 References 195 Modeling and Tracking Dynamic Targets 199 Introduction 199 Target Dynamic Models 200 Singer Acceleration Model 200 White Noise Constant Velocity and Constant Acceleration Dynamic Models 203 Coordinated Turn Models 205 Other Target Dynamic Modeling Approaches 207 Kalman Filter Implementation 208 Horizontal Turn Model With Velocity States 209 Implementation of Pseudomeasurements 21k Nearly Constant Speed Horizontal Turn Model 212 Maneuver Adaptive Filtering Methods 214 Single Filter Reactive Adaptation 214 Variable Dimension Filtering 216 Cascaded Filtering 216 Multiple Model Filtering 221 Interacting Multiple Model Filtering 221 IMM Interaction/Mixing and Prediction 224 Gating and Data Association 225 Likelihood Calculation and Model Probability Update 226 Combining Different State Models 227 Choice of Markov Transition Probabilities 229 Alcernative Mixing Procedures 232 Comparative Performance Study 232 Choice of Methods to Compare 233 Results 235 Three-Dimensional Models 240 Tracking Tactical Ballistic Missile Targets 24] Boost-Stage Missile Dynamics Model 242 Boost-Stage Filter Development 244Contenis xi 4.8.3 Multiple Model Approach 247 4.9 Conclusions 251 References 253 5 Passive Sensor Tracking Methods 259 5.1 Introduction 259 5.2 Initial Orbic Determination 262 5.2.1 Derivation of Laplace Method for IOD 262 5.2.2 Example Results 264 5.2.3 Methods to Reduce IOD Error 267 5.3. State Estimation With Angle-Only Measurements and Ownship Maneuver 268 5.3.1 Cartesian Filtering Approach 269 5.3.2 Modified Polar and Spherical Coordinates Filter Approach 271 5.3.3 Comparative Simulation Results 275 5.4 Tracking With Frequency (Doppler) and Angle Measurement Data 280 5.4.1 Extension of MPC to Frequency/Doppler Measurements 282 5.4.2 Cartesian Coordinate State Estimation Using Angle and Frequency/Doppler Measurements 283 5.5 Use of Measured IR Intensity 286 5.5.1 Time-to-Go Estimation From Amplitude Rate Tracking 286 5.5.2 Single Sensor Multispectral Ranging 290 5.6 Use of Template/Profile Data to Determine ‘Target Type and Boost-Stage Trajectory 293 5.6.1 Overview of Profile Matching Procedure 294 5.6.2 Iterative Search Procedure 299 5.6.3 Use of Measured Intensity 302 5.7 Target Image Tracking (Extended Target Tracking) 303 5.7.1 Algorithm Elements 304 5.7.2 Centroid and Edge Tracking 309 9.7.3 Correlation Tracking 310 5.7.4 Filtering and Data Association 313 5.7.5 Laser Augmented Image Target Tracking 315 5.8 Video Sensor Traffic Monitoring 315 5.8.1 Vehicle Detection 316if Design and Analysis of Modern Tracking Systems 5.8.2 5.8.3 5.9 6.8.1 6.8.2 69 6.9.1 Object Tracking 317 Use of Tracker Output 318 Conclusion 318 References 319 Basic Methods for Data Association 325 Introduction 325 Track Score Function 327 Likelihood Ratio Development 328 Track Score Initiation 330 Special Cases of Signal-Related Data 331 Score-Based Track Confirmation and Deletion 332 Gating 334 Rectangular Gates 335 Ellipsoidal Gates 336 Maneuver Gating 338 Global Nearest Neighbor Method 338 Solution of the Assignment Problem 342 The Auction Algorithm 343 N-Best Solutions to the Assignment Problem 346 The All-Neighbors Data Association Approach 350 The PDA Method 350 Extension to JPDA 353 Combining PDA and IMM 356 PDA Track Initiation and Deletion 357 JPDA Extensions and Modifications 359 Multiple Hypothesis Tracking 360 MHT Tracker Elements 364 Alternative MHT Implementations 365 Node Structure With N-Scan Pruning 367 Presentation of MHT Data 369 GNN, JPDA, and MHT Performance Comparison 369 Comparative Studies in Tracking Literature 370 Simulation Study Results 371 Group Tracking 374 Group Tracking Without Individual Tracks 3746.9.2 6.9.3 6.10 6.10.1 6.10.2 6.10.3 6.10.4 6.10.5 6.11 6.12 6.12.1 6.12.2 6.13 6.13.1 6.13.2 6.13.3 6.14 vA 12 73 73.1 73.2 73.3 7.3.4 7.3.5 74 TAI 7A2 75 Contents Group Tracking Plus Simplified Individual Xiif Tracks 376 Individual Target Tracking Supplemented by Group Information 376 Compensation for Sensor Resolution Limitations 377 Probability of Observation Merging 378 Updated Track Score and State Estimates Using Merged Observations 380 Example Score and Probability Calculation 383 Implementation 385 ‘Warm Start Track Initiation (Track Splitting) 386 Kalman Filter Covariance Increase for Missed Detections 386 Global Nearest Neighbor Enhancements 389 Modified GNN Update 389 Simplified Delayed Decision Logic 389 Choice of Detection Threshold 392 Choice of Global Threshold 392 Adaptive Threshold Setting 393 Summary of Results 394 Conclusions 395 References 396 Advanced Methods for MTT Data Association _403 Introduction 403 The Integer Programming Approach (Morefield’s Method) : 404 Multidimensional Assignment Approach 408 Formulation 408 3D Application of Lagrangian Relaxation 410 Example of Lagrangian Relaxation 414 Extension to N-D Assignment (N> 4) 418 Multiscan Assignment Approach to MHT 422 Dynamic Programming Algorithm 423 Single-Target (Single-Path) Solution 425 Multiple-Target Solution Approach 426 JPDA Extensions 427xiv Design and Analysis of Modern Tracking Systems 75.1 75.2 76 76.1 7.6.2 17 771 772 773 714 78 73 7.9.1 79.2 743 7.10 8.1.3 8.2 6.2.4 8.2.2 8.2.3 84 641 84.2 8.4.3 Mixture Reduction Algorithms 428 Multiple Sean JPDA 431 Augmented State Vector Approach to MTT 433 Coupling Due to Common Target Dynamics and Measurement Error 433 Image Tracking of Crossing Targets 439 State Estimation Without Data Association 440 The Symmetric Measurement Equation Filter 440 Event-Averaged Maximum Likelihood Estimation 445 Batch Processing Maximum Likelihood Estimation Methods 450 Bayesian State Space Estimation 451 Track-Before-Detect 455 Knowledge-Based Methods 458 Neural Network Applications 458 Genetic Algorithms 463 Use of Qualitative/ Heuristic Information 469 Conclusions 473 References 475 Attribute Data Fusion 483 Introduction 483 The Role of Attribute Data Fusion 484 Comments on Selecting an Inferencing Approach 485 Chapter Overview 485 The ED Problem and the Nature of Available A Priori Information 486 A Priori Target ID Modeling 487 ‘Target Attribute Dynamics Modeling 488 Measurement Process Modeling 489 Voting or Set Intersection Techniques for Attribute Data Fusion 490 Classical Statistical Methods 495 Bayesian Inference 496 Maximum A Posteriori Inference and the MAP Techniques 503 Likelihood Inferencing and the Maximem Likelihood Technique 505Contents xv 8.5 Evidential Reasoning for Attribute Data Fusion 507 8.5.1 Evidential Reasoning Primer 509 8.5.2 Implementation of Evidential Reasoning 513 8.5.3. Partial Probability Models 514 8.54 Partial Probability Models and Evidential Reasoning 517 8.5.5 Bayes’s Rule With Parametric Probability Models and Relation to Dempster’s Rule of Combination 521 8.5.6 Partial Transitional Probability Models for Evidential ID Techniques 533 8.5.7 Power Set Approach 544 8.5.8 Typical Set Approach 547 8.5.9 Comparative Example of Bayesian and Fvidential Reasoning for Target ID 551 8.6 More on the Relation Between Evidential Reasoning and Probability 564 8.6.1 Basic Mass Assignment Procedure for Complete Transitional Models 565 8.6.2 Correspondence Between Bayesian and Evidential Reasoning Under Complete Probability Models 567 86.3 Three Prisoner Problem Example Under Complete Probability Models 569 8.6.4 Correspondence Between Bayesian and Eyidential Reasoning Under Complete ‘Transitional and Partial Prior Probability Models 570 8.6.5 Relation of Lower and Upper Probabilities to the Support and Plausibilicy 574 8.6.6 Three Prisoner Problem Example Under Partial Probability Models 575 8.7 Implementing Dempster-Shafer Reasoning 578 8.8 Conclusions 578 References 582 Appendix 8A: Derivation of Maximum and Minimum Bayesian A Posteriori Probabilities Under Partial Prior Distributions 585xvi Design and Analysis of Modern Tracking Systems 9.4 9.2 9.3 9.3.1 9.3.2 9.3.3 9.3.4 9.3.5 9.4.1 9.4.2 9.5 9.5.1 9.5.2 9.5.3 9.6.1 9.6.2 9.6.3 9.6.4 9.6.5 97 9.7.1 9.7.2 9.73 974 9.7.5 9.8 98.1 9.8.2 9.8.3 Multiple Sensor Tracking: Issues and Methods 595 Introduction 595 Basic Principles of Multiple Sensor Tracking System Design 598 Multiple Sensor System Architectures 599 Central-Level Tracking, Centralized Track File 602 Central-Level Tracking, Distributed Track File 607 Sensor-Level Tracking, Centralized Track File 608 Sensor-Level Tracking, Distributed Track File 610 Hybrid System 611 Application Examples 612 Multiple Passive Sensor System Tracking 612 Internetted Multisensor Systems 613 General Expression for Multisensor Data Association 616 Derivation 616 Examples 620 Generalized Assignment Matrix 624 Multisensor Data Association Methods 626 Centralized Architectures 626 Distributed Architectures 627 Overview of Track-to-Track Association 628 Implementations of Track-to-Track Association 629 Extension to MHT and JPDA 637 Distributed Multisensor Detection and ‘Track Initiation 637 Global Detection Test 639 System Detection Performance Optimization 641 Information Transmission to the Fusion Center 643 Other Issues in Multisensor Detection 645 Multiple Sensor Track Confirmation 646 Multiple Sensor Filtering 649 Handling Differing Rates/Out-of-Sequence Data 649 Ensuring That More Is Better 650 Use of Complementary Sensor Data 6509.9 10 10.1 10.2 10.2.1 10.2.2 10.3 10.3.1 10.3.2 10.4 10.4.1 10.4.2 10.4.3 10.5 10.5.1 10.5.2 10.5.3 10.5.4 10.5.5 10.6 10.6.1 10.6.2 10.6.3 10.6.4 10.6.5 10.7 10.7.1 10.7.2 10.8 Contents xvi Conclusions and Further Reading 651 References 654 Multiple Sensor Tracking: System Implementation and Applications 661 Introduction 661 Multiple Sensor Measurement Transformation 662 The Stereographic Coordinate System 663 Nonlinear Measurement Input Form 665 Central-Level Track Filtering 667 Alternative Central-Level Track Updating Methods 667 Filtering Out-of-Sequence Observations: Optimal Filter Development 670 Multiple Sensor Track Fusion 678 Fusion Methods/Error Sources 678 Use of Cross Covariance 681 Use of Equivalent Measurements 684 Multiple Sensor Registration/Misalignment Estimation 689 Registration Error Sources 689 Matching Method to Application 692 Least Squares Estimation Approach 692 Registration Error Estimation via Kalman Filtering 696 Combined Registration and Data Association 697 Distributed Multiple Passive Sensor Systems 699 Choice of Tracking Architecture 700 State Estimation Using Distance of Closest Approach 702 ‘Triangulation Gating and Association Statistics 706 Estimation Accuracy 7U Varad Hinge (Inclination) Angle Method 716 Angle-Only (2D)-to-Position (3D) Track Gating and Fusion 719 Gating and Fusion Relationships 719 Gating and Fusion Example 722 Stereo Track Formation and Maintenance 724Xvili Design and Analysis of Modern Tracking Systems 10.8.1 10.8.2 10.3 "1 11.2.1 11.2.2 11.2.3 113 V4 1 11.6 Wi W721 11.7.2 118 119 11.10 14.41 14.12 12 12.1 12.1.4 12.1.2 Angle-Only Tracking Deghosting Problem 725 Deghosting for Closely Spaced Ballistic ‘Targets 728 Conclusions 729 References 730 Appendix 10A: Derivation of Hinge Angle Statistics 735 Reasoning Schemes for Situation Assessment and Sensor Management 737 Introduction 737 An Introduction to Fuzzy Set Theory and Inexact Reasoning Systems 739 A Fuzzy Set Primer 740 Rule-Based. Systems and Inexact Reasoning 7A2 Implementing Inexact Reasoning 743 Introduction to More Sophisticated Fuzzy Set Theory 746 Some Preliminaries 746 ‘T-Norms, T-Conorms, and Negation Operators 748 The Extension Principle 753 Relations, Composition, and Partial Orderings 756 Fuzzy Relation Basics and Composition 756 Fuzzy Partial Ordering 760 Linguistic Variables and Linguistic Approximation 767 Implication and Inferencing 768 Ranking Objects Using Fuzzy Partial Orderings 716 Fuzzy State Machines 781 Conclusions 794 References 794 Situation Assessment 797 Introduction 797 The Role of Situation Assessment 798 Chapter Organization 801Contents xix 12.2 Domain-Specific Knowledge of Weapon Systems 801 12.2.1 Overview of Tactical Missile Guidance 802 2.2.2 Missile Delivery Methods 802 223 Missile Guidance Laws 804 2.24 Tactical Missile Launch Envelopes 808 12.3 Domain-Specific Knowledge of Tactical Intercepts 8li 2.3.1 Basic Concepts of Tactical Intercepts 812 23.2 Forward Quarter Intercept 814 12.3.3 Stern Conversion Intercept 821 2.3.4 — Variations on the Forward Quarter and the Stern Conversion Intercepts 821 12.4 — Situation Assessment System Architecture 832 2.4.1 Situation Assessment Data Processing Block Diagram 833 12.5 Simple or Static Recognition 837 12.6 Temporal Recognition or Capturing Tactical Intercept Domain-Specific Knowledge 845 12.6.1 Automated Temporal Recognition of Tactical Air Intercepts 847 2.6.2 Temporal Reasoning as a Methodology for Event Fusion 849 12.7 Partial Ordering and Prioritization 852 12.8 Ancillary Situation Assessment Functions 860 12.9 Conclusions 862 References 863 13 Tracking System Performance Prediction and Evaluation 865 13.1 Introduction 865 13.2 Tracking Filter Performance Prediction 866 13.3 Predicting Track Confirmation Performance 871 13.3.1 Limits on Track Confirmation Feasibility 872 13.3.2 Application of SPRT to Track Confirmation 873 13.3.3 SPRT Analysis of Track Confirmation 874 13.3.4 Simplified Monte Carlo Simulation 877 13.35 IRST Example 878 13.4 Analytical Expressions for Predicting Data Association Performance 882