Robust MVDR beamforming based on covariance matrix reconstruction

The minimum variance distortionless response (MVDR) beamformer has better resolution and much better interference rejection capability than the data-independent beamformers. However, the former is more sensitive to errors, such as the steering vector errors caused by direction of arrival mismatch, imprecise array calibration or any other possible factors, especially in the case of high signal-to-noise ratio (SNR). A new robust MVDR beamformer against the general steering vector errors is proposed in this paper. This method is based on the reconstruction of the covariance matrix which aims to reduce the power of the signal of interest (SOI) in the covariance matrix. The eigenanalysis is used in the process of removing the SOI component and modifying the residual covariance matrix. Simulation results show that the proposed method has excellent signal-to-interference-plus-noise ratio (SINR) performance under the mismatch condition.