False alarms in radar detection within sparse-signal processing

Radar-detection metrics are assessed in outcomes of sparse-signal processing (SSP) with test statistics based on the subgradient and the dual feasibility in the SSP optimization via an approach separating false alarms (FAs) from targets. In radar, SSP is aimed for estimating a sparse solution whose FAs are fixed and whose detection of targets is optimal as in traditional detection. Existing detection schemes employing thresholds are based on the theory for a single target and ideal sensing coherence. SSP facilitates development of generic radar-detection metrics by including sensing coherence and multiple targets. We focus on assessing FAs in detection within SSP at different values of sensing coherence, signal-to-noise ratio and a number of targets, and also as compared with the existing detection. The theoretical analysis of detection within SSP is validated with numerical results from range processing.