Noise Source Localization and Optimization of Phased-Array Results

Abstract

A new approach to the deconvolution of acoustic sources has been developed and is presented in this work. The goal of this postprocessing is to simplify the beamforming output by suppressing the side lobes and reducing the sources' main lobes to single (or a small number of) points that accurately identify the noise sources' positions and their actual levels. In this work, a new modeling technique for the beamforming output is proposed. The idea is to use an image-processing-like approach to identify the noise sources from the beamforming maps: that is, recognizing lobe patterns in the beamformed output and relating them to the noise sources that would produce that map. For incoherent sources, the beamforming output is modeled as a superposition of point-spread functions and a linear system is posted. For coherent sources, the beamforming output is modeled as a superposition of complex point-spread functions and a nonlinear system of equations in terms of the sources' amplitudes is posted. As a first approach to solving this difficult problem, the system is solved using a new two-step procedure. In the first step, an approximated linear problem is solved. In the second step, an optimization is performed over the nonzero values obtained in the previous step. The solution to this system of equations renders the sources' positions and amplitudes. In the case of coherent sources, their relative phase can also be recovered. The technique is referred as noise source localization and optimization of phased-array results. A detailed analytical formulation, numerical simulations, and sample experimental results are shown for the proposed postprocessing.