Advanced Sparsity-Driven Models and Methods for Radar Applications

✍ Scribed by Gang Li

• Publisher: SciTech Publishing
• Year: 2020
• Tongue: English
• Leaves: 272
• Category: Library

✦ Synopsis

Compressed sensing (CS) has been one of the most active topics in signal processing
area. By exploiting the sparsity of the signals, CS offers a prospective way for
reducing data amount without compromising the performance of signal recovery or
enhancing resolution without increasing the number of measurements. The signals
in many radar applications are sparse or compressible, so the radar systems may
benefit from the sparsity-driven models and methods in terms of reducing observation
duration, simplifying hardware, and enhancing performance. However, in
practical radar applications, it is found that directly applying the basic CS models
and algorithms to radar data may be less than optimal and even unsatisfactory.
Thus, it is necessary to develop advanced sparsity-based models and algorithms to
fit various radar tasks, which has become a fast-growing branch of radar signal
processing in recent years.
The objective of this book is to introduce more recent developments on
advanced sparsity-driven models and methods that are designed for radar tasks
including clutter suppression, signal detection, radar imaging, target parameter
estimation, and target recognition, mainly based on my publications in the last
decade. Besides the theoretical analysis, numerous simulation examples and
experiments on real radar data are presented throughout the book. The material
presented in this book can be understood by readers who have a fundamental
knowledge of radar signal processing. The book can serve as a reference book for
academic researchers, practicing engineers, and graduate students.
The outline of this book is as follows. Before introducing the advanced sparsitydriven
models and methods designed for radar tasks, the fundamentals of CS are
briefly reviewed in Chapter 1. In Chapter 2, the hybrid greedy pursuit algorithms are
presented for enhancing radar imaging quality. In Chapter 3, the two-level block
sparsity model is introduced to promoting the sparsity of signals of multichannel
radar systems. In Chapter 4, the parametric sparse representation is studied to deal
with model uncertainty during the radar data collection. Chapter 5 investigates how
to simultaneously achieve high-resolution and wide-swath in single-channel synthetic
aperture radar (SAR) imaging by utilizing the Poisson disk sampling.
Chapter 6 concentrates on the sparsity-driven algorithms of radar image formation
from coarsely quantized data. Chapter 7 is concerned with sparsity aware radar
micro-Doppler analysis for micromotion parameter estimation and target recognition.
Chapter 8 is devoted to the distributed detection of sparse signals with
radar networks. Chapter 9 summarizes the book and discusses some perspectives.

✦ Table of Contents

Cover
Contents
About the author
Preface
Acknowledgments
Notation
1 Introduction
1.1 Sparsity of radar signals
1.2 Fundamentals of sparse signal recovery
1.2.1 Signal model
1.2.2 Typical algorithms for sparse signal recovery
1.2.2.1 Regularization methods
1.2.2.2 Greedy algorithms
1.2.2.3 Iterative thresholding algorithms
1.2.2.4 Bayesian algorithms
1.2.3 Beyond the standard sparsity
References
2 Hybrid greedy pursuit algorithms for enhancing radar imaging quality
2.1 Introduction
2.2 Radar imaging with multiple measurement vectors
2.2.1 Signal model
2.2.2 Extended OMP and SP for MMV
2.3 Hybrid matching pursuit algorithm
2.3.1 Analysis of strengths and limitations of OMP and SP
2.3.2 Algorithm description
2.3.3 Computational complexity of HMP
2.3.4 Experimental results
2.4 Look-ahead hybrid matching pursuit algorithm
2.4.1 Algorithm description
2.4.2 The effect of the look-ahead operation
2.4.3 Computational complexity of LAHMP
2.4.4 Experimental results
2.5 Conclusion
References
3 Two-level block sparsity model for multichannel radar signals
3.1 Introduction
3.2 Formulation of the two-level block sparsity model
3.2.1 Clustered sparsity of single-channel data
3.2.2 Joint sparsity of the multichannel data
3.2.3 Two-level block sparsity
3.3 TWRI based on two-level block sparsity
3.3.1 Signal model and algorithm description
3.3.2 Model parameter selection
3.3.3 Computational complexity
3.3.4 Experimental results
3.3.4.1 Simulations
3.3.4.2 Experiments on real radar data
3.4 STAP based on two-level block sparsity
3.4.1 Signal model and algorithm description
3.4.2 Experimental results
3.4.2.1 Simulations
3.4.2.2 Experiments on real radar data
3.5 Conclusion
References
4 Parametric sparse representation for radar imaging with model uncertainty
4.1 Introduction
4.2 Parametric dictionary
4.3 Application to SAR refocusing of moving targets
4.3.1 Problem formulation
4.3.2 Algorithm description
4.3.2.1 Update the sparse solution
4.3.2.2 Update the estimate of the phase compensation parameter
4.3.3 Experimental results
4.3.3.1 Simulations
4.3.3.2 Experiments on real radar data
4.4 Application to SAR motion compensation
4.4.1 Problem formulation
4.4.2 Algorithm description
4.4.2.1 Update the sparse solution
4.4.2.2 Update the motion parameter estimates
4.4.3 Experimental results
4.4.3.1 Simulations
4.4.3.2 Experiments on real radar data
4.5 Application to ISAR imaging of aircrafts
4.5.1 Problem formulation
4.5.2 Algorithm description
4.5.2.1 With known rotation rate of the target
4.5.2.2 With unknown rotation rate of the target
4.5.3 Experimental results
4.6 Conclusion
References
5 Poisson disk sampling for high-resolution and wide-swath SAR imaging
5.1 Introduction
5.2 Tradeoff between high-resolution and wide-swath in SAR imaging
5.3 Poisson disk sampling scheme
5.3.1 Sampling formulation
5.3.2 Comparison between Poisson disk sampling and jittered sampling
5.3.3 Comparison between Poisson disk sampling and random selection from uniform samples
5.4 SAR imaging algorithm with Poisson disk sampled data
5.4.1 Inverse CSA operator
5.4.2 The IST-like algorithm for SAR imaging
5.4.3 Analysis of computational complexity
5.5 Experimental results
5.5.1 Simulations
5.5.2 Experiment on RADARSAT-1 data
5.5.3 Experiment on GAOFEN-3 data
5.6 Conclusion
References
6 When advanced sparse signal models meet coarsely quantized radar data
6.1 Introduction
6.2 Parametric quantized iterative hard thresholding for SAR refocusing of moving targets with coarsely quantized data
6.2.1 Signal model
6.2.2 Description of the PQIHT algorithm
6.2.3 Experimental results
6.2.3.1 Simulations
6.2.3.2 Experiments on real SAR data
6.3 Enhanced 1-bit radar imaging by exploiting two-level block sparsity
6.3.1 Signal model
6.3.2 Description of the E-BIHT algorithm
6.3.2.1 Joint sparsity
6.3.2.2 Clustered sparsity
6.3.2.3 The E-BIHT algorithm
6.3.3 Experimental results
6.3.3.1 Simulations
6.3.3.2 Experiments on real radar data
6.4 Conclusion
References
7 Sparsity aware micro-Doppler analysis for radar target classification
7.1 Introduction
7.2 Micro-Doppler parameter estimation via PSR
7.2.1 Signal model
7.2.2 Description of the POMP algorithm
7.2.3 Discussions
7.2.3.1 Connection with dictionary learning
7.2.3.2 Computational complexity
7.2.3.3 Comparison with Hough-kind algorithms
7.2.4 Simulation results
7.2.4.1 Analysis of parameter estimation accuracy in the noise-free case
7.2.4.2 Analysis of parameter estimation accuracy in noisy environment
7.2.4.3 Analysis of resolution
7.3 Dynamic hand gesture recognition via Gabor– Hausdorff algorithm
7.3.1 Measurements of dynamic hand gestures
7.3.2 Sparsity-driven recognition of hand gestures
7.3.2.1 Extraction of the time–frequency trajectory
7.3.2.2 Clustering for central time–frequency trajectory
7.3.2.3 The nearest neighbor classifier based on modified Hausdorff distance
7.3.3 Experimental results
7.3.3.1 Recognition results with varying sparsity
7.3.3.2 Recognition results with varying time–frequency resolution
7.3.3.3 Recognition results with the varying size of training data
7.3.3.4 Recognition results for unknown personnel targets
7.3.3.5 Analysis of the computational complexity
7.4 Conclusion
References
8 Distributed detection of sparse signals in radar networks via locally most powerful test
8.1 Introduction
8.2 The original LMPT detector
8.2.1 Problem formulation
8.2.2 Formulation of the original LMPT detector
8.2.3 Simulation results
8.3 The quantized LMPT detector
8.3.1 Problem formulation
8.3.2 Formulation of the quantized LMPT detector
8.3.3 Design of the local quantizers
8.3.4 Comparison with the original LMPT detector
8.3.4.1 1-bit quantization
8.3.4.2 Multibit quantization
8.3.5 Simulation results
8.3.5.1 Effect of the designed quantizers
8.3.5.2 Performance of the quantized LMPT detector
8.4 Conclusion
References
9 Summary and perspectives
9.1 Summary
9.1.1 Sparsity-driven radar detection
9.1.2 Sparsity-driven radar imaging
9.1.3 Sparsity-driven radar classification
9.2 Perspectives
9.2.1 Structured models
9.2.2 Practical databases
9.2.3 Combination of sparsity and data-driven algorithms
9.2.4 Generalized sparsity of heterogeneous data
References
Index
Back Cover