Steering vector non-identifiability in covariance matrix fitting based beamforming

Over the last decade, several robust adaptive receive beamformers based on covariance matrix fitting (CMF) techniques have been proposed. For these beamformers, the estimate of the signal steering vector is first matched to the sample covariance matrix. Subsequently, the minimum variance distortionless response beamformer weight vector is computed using the matched vector. The CMF problems are formulated using uncertainty sets for the signal steering vector. However, the signal steering vector is typically non-identifiable if only an uncertainty set for this vector and the array covariance matrix are known. In this paper, we numerically analyze the impact of this non-identifiability on the performance of CMF based beamformers.