Robust adaptive beamforming for passive sonar using eigenvector/beam association and excision

Abstract

A new eigenvector-based robust adaptive beamforming (ABF) algorithm is presented for use with passive sonar arrays. Robustness refers to protection against self-nulling in the presence of signal mismatch. The algorithm is based on an eigenvector decomposition, similar to ABF algorithms such as dominant mode rejection (DMR) and principal components inverse (PCI). For robustness, the proposed method removes the eigenvector from the ABF weight computation deemed as the contributor to self-nulling. This approach is in contrast to eigenvector-based ABF with a white noise gain constraint which instead changes the weighting of all eigenvectors. By identifying and removing the offending eigenvector, robustness to self-nulling can be restored. The proposed eigenvector/beam association and excision (EBAE) algorithm, associates each eigenvector with a beam or angle and removes it from that beam only. The EBAE eigenvector-based ABF is compared to other robust ABF algorithms in both simulation and experimental data where it is shown to provide improved performance both in terms of interference rejection and signal separation.