A cross-relation based matched field processor for source localization with random noise sources

Abstract

This paper presents a cross-relation (CR) based matched field processing (MFP) technique for source localization in a shallow water environment where the source propagates a random signal. The estimation formulas are given for nonstationary (NS) and wide sense stationary (WSS) random sources. For each case two formulations are proposed, a self-CR and a cross-CR according to which channel output signal is used to construct the estimator. We demonstrate the performance of the proposed estimators in source localization, and make comparison with two common estimators, the Bartlett and minimum variance (or Capon's) estimators. The comparison is carried out first by simulation using wide band, WSS source noise in a shallow water environment whose represents the continental shelf offshore Vancouver Island. Subsequently, comparison uses real ship noise obtained in an experiment in the same region. The simulation results show that in the low signal-to-noise ratio (SNR) the cross-CR based estimator gives superior performance compared to the self-CR, Bartlett and MV estimator with respect to resolution and side lobe level. For real ship noise both CR based MFPs show similar results in comparison with Bartlett processor, i.e. lower side lobe levels throughout the ambiguity surface and narrower main lobe around source locations in comparison to what we have for the Bartlett processor.