Overcoming snapshot-deficient measurements with knowledge-aided approaches.

Abstract

The use of knowledge-aided covariance is considered for processing underwater acoustic array data in snapshot-deficient scenarios. The knowledge-aided formalism is a technique that combines array data with a known covariance to produce an invertible estimate. For underwater acoustics, simulations of ambient noise provide the a priori covariance allowing degraded signals to be processed adaptively in situations where the sample covariance matrix is rank-deficient. The method is demonstrated for matched field processing using the 21 element array event S5 from the SWellEx-96 experiment. With five snapshots, the knowledge-aided approach significantly reduces localization ambiguity compared to the adaptive white noise gain constraint processor.