Deconvolutional beamforming for air and underwater acoustic sensor arrays

Abstract

Autonomous acoustic sensor arrays are often plagued with poor spatial resolution capabilities, especially at the lowest frequencies. Beamforming techniques that are data-adaptive can improve upon the resolution capability of conventional beamformers under certain conditions. An alternate nonadaptive technique that is simply implemented and that achieves sidelobe reduction while retaining or narrowing main lobe beamwidth is presented. The method for weight matrix construction is based on regularized, constrained deconvolution. These deconvolutional beamformer (DBF) beamspace weights, which are precomputed off-line using this nonlinear process, are applied to the data using a linear process. The physical array may be smaller when using the DBF since this beamformer simultaneously reduces the width of the main lobe at low frequencies while reducing the integrated sidelobe level. This is something even ‘‘optimal’’ beamformers have difficulty doing since the main lobe generally increases as the sidelobes are reduced. One additional DBF attribute, when compared to most traditional weighting functions, is its compatibility with nonuniformly spaced arrays, including the special case of receive elements that are known to be missing. The automated procedure for determining these beamspace weights is discussed and results are shown for both air and underwater acoustic array data.