Modified Eigenspace Based Iterative Robust Capon Beamformer

Abstract

To improve the performance of adaptive beamformer with large steering vector mismatch, a modified eigenspace based iterative robust capon beamformer (ME-IRCB) is proposed. The approach iteratively implements the robust Capon beamformer (RCB) with an adaptively adjusted uncertainty set. Based on the modified eigenspace method, the desired signal subspace is given, whereafter the size of the uncertainty set is adaptively adjusted according to the mismatch between the estimated and presumed steering vector of the desired signal at each iteration. The proposed approach gives two estimation methods about the uncertainty size, and both of them contribute to the robustness improving of the beamformer. Moreover, unlike other robust beamformers, the proposed approach does not take the integration of the steering vector’s outer product over the adjacent direction of the desired signal, and it works well with the adjacent directions of the desired signal and interference. This merit is due to that, the proposed approach only uses the fundamental sample covariance matrix during the subspace construction, therefore the influence of imprecise steering vector on the subspace is overcame. Simulation results show that the proposed approach performs better than the existing robust beamformers under different evaluation criteria.