Robust matched field-processing algorithm based on feature extraction

Abstract

In this paper, a new matched field processing (MFP) method is presented. The new method is based on the eigenvector estimation, which is also called feature extraction in pattern recognition. It aims at gaining robustness for the environmental mismatch and simultaneously reducing the time-consuming procedure related to forward modeling. To achieve this goal, common signals (eigenvectors) about the assumed replicas are extracted in the process of randomization over environmental parameters. Discarding these eigenvectors and forming an incoherent subspace spanned by the remaining eigenvectors, the source positions can be estimated correctly by a processor described in this paper. The robustness and the effectiveness of the suggested algorithm will be illustrated through the numerical simulations when there are uncertainties in the environmental parameters of the acoustic waveguide.