Single-snapshot beamformer performance using large arrays with faulty sensors

Large sensor arrays are prone to element failures which can seriously degrade their performance. In the event of sensor failures, conventional non-adaptive re-design of array weights for the resulting non-uniform or even sparse array often leads to undesirable mainlobe-versus-sidelobe level trade-offs. Alternatively, adaptive array processing techniques can be employed but for large arrays it can be compromised by lack of snapshot support in dynamic environments. Recently, an approach called beamspace adaptive channel compensation (BACC) was introduced based on the idea of reconstructing the filled array data from the receive beams in which the sidelobe leakage of strong interferers from adjacent beams is minimized. In this paper, BACC and various alternatives are evaluated using actual radar data with both injected and real scatterers. Results of four methods, BACC, MV adaptive beamforming using an augmented Toeplitz covariance matrix (Toeplitz-MV), principal solution beamforming, and conventional beamforming in the presence of faulty sensors are compared in terms of ROC performance, probability of detection versus SINR, and the unmasking of targets in estimated range-Doppler spectra.