High Resolution Detection and Localization of Targets in Noisy Environment of Compressed Sensing Radar

Abstract

AbstractUsing Compressed Sensing (CS), any sparse signal can be measured with far fewer samples than required by Nyquist sampling. Storage space and high-rate ADC requirement can be greatly reduced with CS when the signal is noise-free. Generally, any signal will have noise inherently. When the echo signal received is noisy, traditional Matched Filter (MF) could better optimize noise than CS with optimization algorithms Basis Pursuit Denoising (BPDN) and Least Absolute Shrinkage and Selection Operator (LASSO). To improve the performance of CS with BPDN & LASSO, more measurements are to be taken which leads to an increase in storage space and high-rate ADC requirement than that of a noise-free case. To overcome this, Forward Stepwise Regression using False Discovery Rate (SW-FDR) stopping rule is utilized for detection and range estimation of stationary multiple point targets. SW-FDR could effectively optimize noise that leads to a reduction in storage space and high-rate ADC requirement drastically even in presence of noise.KeywordsCompressed sensingMatched FilterBPDNLASSOSW-FDR