Enhanced Beamforming via Bayesian Postprocessing

A Bayesian postprocessing (BP) scheme for improving beamforming performance is proposed. It is implemented as a signal beam power estimator at a beamformer output based on the Bayesian minimum mean square error (BMMSE) criterion, boosting signal components by suppressing noise and interferences. The BMMSE estimation of the signal beam power boils down to the track of the noise-plus-interference power and the signal-to-interference-plus-noise ratio (SINR), which are achieved with the minimum-statistics (MS) method and the decision-directed approach (DDA), respectively. The Bayesian postprocessing scheme is applied to the conventional beamforming (CBF), leading to the BP-CBF. Simulation results showed the BP-CBF significantly outperforms the CBF without sacrificing its robustness to signal mismatch.