Target Detection in Passive Radar Sensors Using Least Angle Regression

Passive bistatic radars (PBRs) use illuminators of opportunity to detect and localize targets. Exploiting signals of these sources which are not designed for radar applications results in essential challenges in target detection, and requires special signal processing techniques. In this paper, we propose a new approach for target detection in PBRs by formulating the problem as a linear regression. To solve this problem, we take advantage of the sparsity of received signals in the range-Doppler domain which enables us to employ statistical model selection algorithms, such as LASSO or LAR. In contrast to the most existing PBR algorithms, the proposed method does not require to specify a prior subspace for clutter and eliminate interferences before target detection. This advantage is achieved because our algorithm identifies targets, clutter, and direct-path simultaneously within a unified procedure. Our extensive simulation results illustrate that the proposed method performs very close to the optimal upper band performance (i.e., that of the matched-filter based detector) in the single-target scenario. Moreover, our results reveal that our algorithm has high detection performance in multitarget scenarios with the presence of interfering targets, strong clutter, and a very powerful direct-path.