A Bayesian Robust Adaptive Beamforming Based on Phase-Invariant MMSE Criterion

The Bayesian beamformer under minimum mean square error (MMSE) criterion has been reported to perform well against uncertainty in source direction of arrival (DOA), but its performance depends on the location of phase reference. This letter proposes a phase-invariant MMSE criterion under which the Bayesian beamformer is independent of the location of phase reference. An iterative algorithm is developed for designing such a new Bayesian beamformer. Numerical examples demonstrate that the new Bayesian beamformer can provide better performance than the existing Bayesian MMSE beamformer for either uniform or non-uniform linear array.