Research on the robustness of array beamforming based on Support Vector Regression

In practical array systems, traditional adaptive beamforming algorithms will degrade if some unexpected assumptions become wrong or imprecise. Therefore, the robustness of adaptive beamforming techniques against environmental and array imperfections is one of the key issues. Compared with traditional methods, robust beamforming brings about much improvement on array performance. In this paper, we propose a new approach to adaptive beamforming that provides increased robustness against the mismatch problems. After generalizing the conventional linearly constrained minimum variance cost function by including a regularization term to penalize differences between the actual and the ideal array responses, we adopt the form of Support Vector Regression (SVR) to increase the beamformer robustness against errors by utilizing ε -insensitive loss function for the penalty term. In comparison with other robust beamforming techniques especially for high SNR scenarios, simulation results show that the proposed SVR-based beamformer has the desired robust performance both in no-mismatch and mismatch situations regardless of the number of interference signals.