Minimum Output Energy Beamforming Based on Support Vector Regression

We propose a beamforming method based on Support Vector Regression (SVR) for uniform linear arrays (ULAs). In the proposed algorithm, a diagonal value is added to the covariance matrix of the structural risk item, to ensure the matrix invertible. Moreover, the proposed method not only carries out the minimum output energy, but also averts the low robustness caused by the direction of arrival mismatch and the limited number of snapshots. Performance of the proposed algorithm is evaluated by numerical simulations, which is compared with the minimum variance distortionless response (MVDR). It is illustrated that the SVR-based algorithm performs better than MVDR with small samples and high signal-to-noise ratio (SNR) scenarios.