Matched-field Source Localization via Statistical Covariance Matching

Matched-field methods concern estimation of source locations and/or ocean environmental parameters by exploiting full wave modeling of acoustic waveguide propagation. Typical estimation performance demonstrates that sidelobe ambiguities dominate the estimation at low signal-to-noise ratio (SNR), leading to a threshold performance behavior. An adaptive estimator, such as minimum variance distortionless response (MVDR), has the estimate-subtract structure, which can lead to the sidelobe suppression. However, in practical circumstances, the lack of snapshots would cause dramatic performance deterioration of MVDR. In this paper, we bring in a new matched-covariance estimator (MCE) into matched-field processing (MFP) to attack such weakness. Performance analysis of MCE is implemented for source localization in a typical shallow water environment chosen from the 2001 Asian Seas International Acoustic Experiment (ASIAEX). Simulation results demonstrate that 1) MCE owns the resolution benefit of an adaptive method while being robust against snapshot deficiency; 2) in terms of MSE, it performs almost equally well with the maximum likelihood estimator (MLE) and can outperform the MVDR; 3) for the threshold SNR concern, MCE has a threshold SNR close to that of the MLE and much lower than that of the MVDR. Overall, MCE achieves a trade-off between spatial resolution and snapshot-number robustness.