Robust Adaptive Beamforming using Desired Signal Steering Vector Estimation and Variable Loading

Abstract

It is known that the desired signal steering vector error, the number of signal samples and the input signal to noise ratio (SNR) are crucial factors to the adaptive beamforming performance. In the presence of steering vector mismatch or lack of samples or strong desired signal, the minimum variance distortionless response (MVDR) beamformer can generate the distorted mainbeam with high sidelobe levels. Diagonal loading of the sample covariance matrix is a widespread technique to provide robustness against such cases. However, there is a tradeoff between the robustness improvement and interference and noise cancellation capability to determine a proper value of loading. Rather than a fixed loading as the diagonal loading, variable loading can provide more robust and protect the rising sidelobe levels in the presence of mismatch. To remedy the effect of mismatch, the presumed desired signal steering vector is utilized to estimate its actual one. The estimation is done by the max/min optimization of the array output power. Then, an algorithm to create the robust MVDR beamformer against the desired signal steering vector mismatch is presented by using the estimated desired signal steering vector and variable loading. Simulation results show that the proposed method has significantly beampattern improvement when the error due to the steering vector mismatch, small number of signal samples and high input SNR exist.