Ambiguity analysis of matched-field source localization under spatially correlated noise field

Abstract

Matched-field processing (MFP) concerns estimation of source location by exploiting full-wave modeling of acoustic waveguide propagation. The ambiguity output in MFP relies on the correlation between the signal field at the true source position and the signal field at each scanning source position. This correlation often shows a multimodal structure due to the nonlinear parameter dependence: in addition to a mainlobe around the true source position, there are unpredictable prominent high sidelobes elsewhere. For a well-conditioned problem, in the absence of noise the peak output is guaranteed to occur at the true source position. In the presence of noise, we will probably have some peak outputs around the sidelobes, introducing a large localization error. Therefore, output ambiguity structure is a very important factor in the development of any matched-field algorithm operated in low-SNR scenarios. To analyze the ambiguity behavior, a quantitative approach for error analysis has previously been developed in the context of the maximum likelihood estimate (MLE) with spatially white noise. In this talk, generalization to including spatially correlated noise field introduced by discrete interferences and/or surface distribution sources is discussed.