Model-based space-time adaptive processing for active sonar

In this paper, the problem of partially adaptive space-time adaptive processing (STAP) for towed active sonar arrays operating in a strong reverberation environment under extremely snapshot deficient conditions is considered. In particular, we propose a robust STAP technique, which incorporates a physical model for the space-time structure of the received reverberation across range, in order to form a model-based covariance matrix, used to regularize the sample covariance matrix while computing the adaptive weights. Simulation as well as real data results using a bistatic towed active sonar array are presented which demonstrate the improved detection performance obtained using our technique compared to conventional space-time processing and other partially adaptive schemes.