Underwater source localisation utilising interference pattern under low SNR conditions

Abstract

Underwater acoustic source localisation in deep water shadow zone is difficult since the vertical aperture of the array is limited and the received signal-noise ratio (SNR) is low. This article investigates a novel interference pattern in the first shadow zone. The intensity displays two oscillating striations in the interference pattern, which can be used to estimate the source depth under the restriction of lacking adequate spatial samples and acoustic environmental parameters, and the range of the source can be estimated roughly using the position of which. The method is applicable to multi-target localisation. In practice, the interference pattern is always corrupted by noise under low SNR conditions, resulting in performance degradation or even failure of the localisation. To address the problem of recovering the corrupted interference pattern, an interference tensor is constructed and analysed, which expands the general two-dimensional interference pattern into range-grazing angle-frequency domain to make use of the multi-frame interaction coherently. A Tucker decomposition-based method that adopts a low-rank representation of the interference tensor is proposed to retrieve the intact interference pattern. Simulated data validate the accuracy of the localisation under low SNR conditions. It is shown that this method significantly improves the estimated performance over conventional matched-field processing.