Constant false alarm rate detection processes are important in radar signal processing. Such detection strategies are used as an alternative to optimal Neyman-Pearson based decision rules, since they can be implemented as a sliding window process running on a radar range-Doppler map. This book examines the development of such detectors in a modern framework. With a particular focus on high resolution X-band maritime surveillance radar, recent approaches are outlined and examined. Performance is assessed when the detectors are run in real X-band radar clutter. The book introduces relevant mathematical tools to allow the reader to understand the development, and follow its implementation.
Graham Weinberg (Defence Science and Technology Group (DSTG) Australia)
Country of Publication:
14 September 2017
SECTION I: PRELIMINARIES Introduction Purpose Sliding Window Detectors Range-Time Intensity Example Historical Development Mathematical Formulation Detectors in Exponentially Distributed Clutter Some Fundamental Concepts Structure of the Book Probability and Distribution Theory for Radar Detection Outline Fundamentals of Probability Transformations Moments Inequalities Jointly Distributed Random Variables Conditional Distributions Some Special Functions of Random Variables Order Statistics Uniform Distributions and Simulation Properties of Estimators Spherically Invariant Random Processes Distributions for X-Band Maritime Surveillance Radar Clutter Introduction Early Models for Clutter The Weibull Distribution K-Distribution The Pareto Class of Distributions Pareto Type Distributions Properties of the Pareto Distribution Parameter Estimation Pareto Model Adopted for Detector Development SECTION II: FUNDAMENTAL DETECTION PROCESSES Adaptation of Exponential Detectors to Pareto Type I Distributed Clutter Introduction General Considerations The Order Statistic Detector The Cell-Averaging Detector The Geometric Mean Detector Performance in Homogeneous Clutter Effect of Interfering Targets Clutter Transitions Conclusions A Transformation Approach for Radar Detector Design Introduction The Transformation Approach Examples of Detector Performance Preservation of the CFAR Property Lomax-Distributed Clutter and Detector Performance Modification of the General Transformed Detector Specialisation to the Pareto Clutter Case Performance of the New CFAR Processes Modified Minimum Order Statistic Detector Introduction Transformed Order Statistic Detectors Detector Comparison Mathematical Analysis of Detectors Selection of Parameter M Performance in Homogeneous Clutter Examples of Management of Interference False Alarm Regulation Conclusions Dual Order Statistic CFAR Detectors Introduction Motivation and Definition of Detection Process Specialisation to the Pareto Type I Case Performance in Homogeneous Clutter Management of Interfering Targets False Alarm Regulation Conclusions On Goldstein's Log-t Detector Introduction The Log-t Detector An Order Statistic Based Log-t Detector Performance in Homogeneous Clutter Interference False Alarm Regulation Conclusions SECTION III: SPECIALISED DEVELOPMENTS Switching Based Detectors Introduction Development of a Switching Detector Generalisation of the Switching Detector Switching CFAR Detector Performance of the SW-CFAR Detector Conclusions Developments in Binary Integration Introduction Binary Integration Mathematical Analysis of Binary Integration Binary Integration Parameter S Performance in Homogeneous Clutter Performance with Interference Clutter Transitions Conclusions Detection in Range Correlated Clutter Introduction Decision Rule in Correlated Clutter Mardia's Multivariate Pareto Model Order Statistic Decision Rule Thresholds Performance Analysis Analysis of the Minimum-Based Detector Achieving CFAR in Correlated Pareto Distributed Clutter Conclusions SECTION IV: FURTHER CONCEPTS Invariance and the CFAR Property Introduction Group Theory Basics The Invariance Property Some Invariant Statistics Examples of Invariant CFAR Detectors Performance of Invariant Detectors Conclusions Convergence and Approximation of the Pareto Model Introduction Problem Specification Information Theory Kullback-Leibler Divergence Conclusions Appendices References Index
Graham V. Weinberg completed his B.S. and Ph.D. degrees at the University of Melbourne, Australia. His doctoral thesis examined distributional approximations of stochastic processes using the Stein-Chen method. After a short period in telecommunications research at the University of Adelaide, he joined Defence Science and Technology Group, Australia. In the capacity of a scientist, he has undertaken research into radar detection issues arising from airborne high resolution X-band maritime surveillance platforms. To further continue his professional development, he has also completed a Master's degree in signal and information processing through the University of Adelaide, Australia. His research interests include CFAR, coherent multi-look radar detection and the mathematics of radar signal processing. He has published extensively and is a member of the Institution of Engineering and Technology (IET), UK.