SIGNAL SUBSPACE BEAMFORMING IN THE PRESENCE OF COHERENT INTERFERENCESAND UNKNOWN CORRELATED NOISE FIELDS

In a realistic passive sonar scenario, the desired signal will be rece ived in the presence of coherent interferences and spatially correlate d noise fields. In this paper beamforming algorithms are developed for this scenario. Two types of signal subspace beamformers based on diff erent optimality criteria, are developed. They are (i) optimum-interfe rence notch beamformer, and (ii) optimum-interference plus noise rejec tor, The beamformers are first developed for the narrowband data model and then extended to the wideband case. Projection matrices onto the signal subspace are computed by using Q-R decomposition instead of the computationally intensive eigen-decomposition. Expressions for the ou tput-signal to interference-plus-noise ratio are derived and used as a measure to evaluate the performance of the beamformers. Simulation re sults are also given to support the theory. (C) 1997 Elsevier Science B.V.