Towed array beamforming using sparse Bayesian learning during maneuvering

A large aperture array is widely used to detect the silent targets among loud interferences. During maneuvering, it is critical to determine the array shape. The array can be modeled as a parabola array during slow maneuvering, thus the hydrophones positions are estimated using one parameter – the bow of the parabola. A sub-array method is proposed to estimate the bow which is consistent with both non-acoustical and acoustical means. Sparse Bayesian learning (SBL) algorithm is then exploited to estimate the directions of arrival (DOAs) using a long aperture array involving the straight and curved shape. The results demonstrate that SBL has better abilities in detecting weak sources among strong sources at low frequencies. During maneuvering, the mirror sources are suppressed thus the left-right ambiguity is resolved. Consequently, SBL can be well applied in passive Sonar systems.