Cochran / Kastella / Castanon Foundations and Applications of Sensor Management

Signals and Communication Technology
1. Auflage 2007
ISBN: 978-0-387-49819-5
Verlag: Springer-Verlag
Format: PDF
Kopierschutz: Adobe DRM (»Systemvoraussetzungen)

E-Book, Englisch, 310 Seiten

This book covers control theory signal processing and relevant applications in a unified manner. It introduces the area, takes stock of advances, and describes open problems and challenges in order to advance the field. The editors and contributors to this book are pioneers in the area of active sensing and sensor management, and represent the diverse communities that are targeted.

Autoren/Hrsg.

Cochran, Douglas

Kastella, Keith

Castanon, David

Cochran, Doug

Inhaltsverzeichnis

1;Preface;6
1.1;Acknowledgments;7
2;Contents;8
3;Contributing Authors;13
4;Symbol Index;15
5;OVERVIEW OF BOOK;17
5.1;1. Introduction;17
5.2;2. Scope of Book;18
5.3;3. Book Organization;19
6;STOCHASTIC CONTROL THEORY FOR SENSOR MANAGEMENT;22
6.1;1. Introduction;22
6.2;2. Markov Decision Problems;25
6.2.1;2.1 Dynamic Programming;27
6.2.2;2.2 Stationary Problems;28
6.2.3;2.3 Algorithms for MDPs;33
6.3;3. Partially Observed Markov Decision Problems;34
6.3.1;3.1 MDP Representation of POMDPs;36
6.3.2;3.2 Dynamic Programming for POMDPs;39
6.4;4. Approximate Dynamic Programming;41
6.5;5. Example;42
6.6;6. Conclusion;47
7;INFORMATION THEORETIC APPROACHES TO SENSOR MANAGEMENT;48
7.1;1. Introduction;48
7.2;2. Background;50
7.2.1;2.1 a-Entropy, a-Conditional Entropy, and a-Divergence;51
7.2.2;2.2 Relations Between Information Divergence and Risk;53
7.2.3;2.3 Fisher Information and Information Divergence;55
7.3;3. Information-Optimal Policy Search;55
7.4;4. Information Gain Via Classification Reduction;58
7.5;5. A Near Universal Proxy;59
7.6;6. Information Theoretic Sensor Management for Multi- target Tracking;62
7.6.1;6.1 The Model Multi-target Tracking Problem;63
7.6.2;6.2 R ´ enyi Divergence for Sensor Scheduling;64
7.6.3;6.3 Multi-target Tracking Experiment;65
7.6.4;6.4 On the Choice of;65
7.6.5;6.5 Sensitivity to Model Mismatch;66
7.6.6;6.6 Information Gain vs Entropy Reduction;67
7.7;7. Terrain Classification in Hyperspectral Satellite Imagery;68
7.7.1;7.1 Optimal Waveform Selection;69
7.8;8. Conclusion and Perspectives;72
8;JOINT MULTI-TARGET PARTICLE FILTERING;73
8.1;1. Introduction;73
8.2;2. The Joint Multi-target Probability Density;76
8.2.1;2.1 General Bayesian Filtering;78
8.2.2;2.2 Non-Linear Bayesian Filtering for a Single Target;79
8.2.3;2.3 Accounting for Target Birth and Death;80
8.2.4;2.4 Computing Renyi Divergence;81
8.2.5;2.5 Sensor Modeling;82
8.3;3. Particle Filter Implementation of JMPD;85
8.3.1;3.1 The Single Target Particle Filter;86
8.3.2;3.2 The Multi-target Particle Filter;87
8.3.3;3.3 Permutation Symmetry and Improved Importance Densities for JMPD;88
8.3.4;3.4 Multi-target Particle Proposal Via Individual Target Proposals;89
8.3.5;3.5 Multi-target Particle Proposal Via Joint Sampling;93
8.3.6;3.6 Partition Ordering;95
8.3.7;3.7 Estimation;97
8.3.8;3.8 Resampling;99
8.4;4. Multi-target Tracking Experiments;99
8.4.1;4.1 Adaptive Proposal Results;100
8.4.2;4.2 Partition Swapping;103
8.4.3;4.3 The Value of Not Thresholding;103
8.4.4;4.4 Unknown Number of Targets;104
8.5;5. Conclusions;105
9;POMDP APPROXIMATION USING SIMULATION AND HEURISTICS;108
9.1;1. Introduction;108
9.2;2. Motivating Example;110
9.3;3. Basic Principle: Q-value Approximation;111
9.3.1;3.1 Optimal Policy;111
9.3.2;3.2 Q-values;112
9.3.3;3.3 Stationary policies;113
9.3.4;3.4 Receding horizon;113
9.3.5;3.5 Approximating Q-values;113
9.4;4. Control Architecture;114
9.4.1;4.1 Controller;115
9.4.2;4.2 Measurement filter;115
9.4.3;4.3 Action selector;116
9.5;5. Q-value Approximation Methods;117
9.5.1;5.1 Basic approach;117
9.5.2;5.2 Monte Carlo sampling;117
9.5.3;5.3 Relaxation of optimization problem;118
9.5.4;5.4 Heuristic approximation;119
9.5.5;5.5 Parametric approximation;120
9.5.6;5.6 Action-sequence approximations;122
9.5.7;5.7 Rollout;123
9.5.8;5.8 Parallel rollout;124
9.5.9;5.9 Control architecture in the Monte Carlo case;124
9.5.10;5.10 Belief-state simplification;127
9.5.11;5.11 Reward surrogation;128
9.6;6. Simulation Result;129
9.7;7. Summary and Discussion;131
10;MULTI-ARMED BANDIT PROBLEMS;133
10.1;1. Introduction;133
10.2;2. The Classical Multi-armed Bandit;134
10.2.1;2.1 Problem Formulation;135
10.2.2;2.2 On Forward Induction;137
10.2.3;2.3 Key Features of the Classical MAB Problem and the Nature of its Solution;139
10.2.4;2.4 Computational Issues;142
10.3;3. Variants of the Multi-armed Bandit Problem;146
10.3.1;3.1 Superprocesses;146
10.3.2;3.2 Arm-acquiring Bandits;149
10.3.3;3.3 Switching Penalties;150
10.3.4;3.4 Multiple Plays;152
10.3.5;3.5 Restless Bandits;154
10.3.6;3.6 Discussion;159
10.4;4. Example;160
10.5;5. Chapter Summary;163
11;APPLICATION OF MULTI-ARMED BANDITS TO SENSOR MANAGEMENT;164
11.1;1. Motivating Application and Overview ;164
11.1.1;1.1 Introduction;164
11.1.2;1.2 SM Example of Multi-armed Bandit;165
11.1.3;1.3 Organization and Notation of This Chapter;166
11.2;2. Application to Sensor Management;166
11.2.1;2.1 Application of the Classical MAB;167
11.2.2;2.2 Single Sensor with Multiple Modes;170
11.2.3;2.3 Detecting New Targets;171
11.2.4;2.4 Sensor Switching Delays;171
11.2.5;2.5 Multiple Sensors;172
11.2.6;2.6 Application of Restless Bandits;173
11.3;3. Example Application;173
11.3.1;3.1 MAB Formulation of SM Tracking Problem;174
11.3.2;3.2 Index Rule Solution of MAB;175
11.3.3;3.3 Numerical Results and Comparison to Other Solutions;177
11.4;4. Summary and Discussion;184
12;ACTIVE LEARNING AND SAMPLING;187
12.1;1. Introduction;187
12.1.1;1.1 Some Motivation Examples;188
12.2;2. A Simple One-dimensional Problem;189
12.3;3. Beyond 1d - Piecewise Constant Function Estimation;200
12.4;4. Final Remarks and Open Questions;209
13;PLAN-IN-ADVANCE ACTIVE LEARNING OF CLASSIFIERS;211
13.1;1. Introduction;211
13.2;2. Analytical Forms of the Classifier;213
13.3;3. Pre-labeling Selection of Basis Functions;214
13.4;4. Pre-labeling Selection of Data;219
13.5;5. Connection to Theory of Optimal Experiments;220
13.6;6. Application to UXO Detection;222
13.6.1;6.1 Magnetometer and electromagnetic induction sensors;223
13.6.2;6.2 Measured sensor data from the Jefferson Proving Ground;223
13.6.3;6.3 Detection results;224
13.7;7. Chapter Summary;229
14;APPLICATION OF SENSOR SCHEDULING CONCEPTS TO RADAR;231
14.1;1. Introduction;231
14.2;2. Basic Radar;232
14.2.1;2.1 Range-Doppler Ambiguity;232
14.2.2;2.2 Elevation and Azimuth;236
14.2.3;2.3 Doppler Processing;239
14.2.4;2.4 Effects of Ambiguity;242
14.3;3. Measurement in Radar;243
14.4;4. Basic Scheduling of Waveforms in Target Tracking;244
14.4.1;4.1 Measurement Validation;245
14.4.2;4.2 IPDA Tracker;245
14.5;5. Measures of Effectiveness for Waveforms;249
14.5.1;5.1 Single Noise Covariance Model;250
14.5.2;5.2 Integrated Clutter Measure;250
14.5.3;5.3 Approximation of ICM;251
14.5.4;5.4 Simulation Results;253
14.6;6. Scheduling of Beam Steering and Waveforms;255
14.6.1;6.1 Tracking of Multiple Maneuvering Targets;255
14.6.2;6.2 Scheduling;256
14.6.3;6.3 Simulation results;258
14.7;7. Waveform Libraries;260
14.7.1;7.1 LFM Waveform Library;263
14.7.2;7.2 LFM-Rotation Library;264
14.8;8. Conclusion;265
15;DEFENSE APPLICATIONS;267
15.1;1. Introduction;267
15.2;2. Background;269
15.3;3. The Contemporary Situation;270
15.4;4. Dynamic Tactical Targeting (DTT);272
15.5;5. Conclusion;276
16;APPENDICES;279
16.1;1. Information Theory;279
16.1.1;1.1 Entropy and Conditional Entropy;279
16.1.2;1.2 Information Divergence;281
16.1.3;1.3 Shannon’s Data Processing Theorem;281
16.1.4;1.4 Shannon Mutual Information;282
16.1.5;1.5 Further Reading;282
16.2;2. Markov Processes;283
16.2.1;2.1 Definition of Markov Process;283
16.2.2;2.2 State-transition Probability;284
16.2.3;2.3 Chapman-Kolmogorov Equation;285
16.2.4;2.4 Markov reward processes;285
16.2.5;2.5 Partially Observable Markov Processes;286
16.2.6;2.6 Further Reading;288
16.3;3. Stopping Times;288
16.3.1;3.1 Definitions;288
16.3.2;3.2 Example;289
16.3.3;3.3 Stopping Times for Multi-armed Bandit Problems;290
16.3.4;3.4 Further Reading;291
17;References;292
18;Index;313