Improving ray-based blind deconvolution of random shipping sources with short arrays in an ocean waveguide using adaptive beamforming

Abstract

This paper introduces an improved ray-based blind deconvolution(RBD) algorithm for sources of opportunity, such as shipping noise, recorded on a short vertical line array (VLA) by using adaptive beamforming. The original RBD algorithm [Sabra et al., JASA, 2010, EL42-7] relies on first estimating the unknown phase of the random source using broadband beamforming along a well-resolved ray path to estimate the full channel impulse responses (CIR) between the unknown source and the VLA elements (up to an arbitrary time-shift) as well as recovering the radiated signal by the random source. Hence, RBD performance is limited by the VLA ability to separate adjacent ray path, especially for the case of small aperture. To address this limitation, broadband MVDR is used as a high resolution estimator to improve the RBD performance when using single snapshot recordings and short VLAs. To do so, smoothing techniques are used to increase the rank of cross-spectral data matrix since the ray arrivals emanating from the same random source are naturally correlated in an ocean waveguide when using single snapshot recordings. The improvement of this proposed RBD algorithm will be demonstrated using both numerical simulation and at-sea shipping noise data recorded in shallow water.