Robust adaptive beamforming using variable loading

It is well known that the performance of adaptive beamformers may degrade in the presence of steering errors or lack of training data. Diagonal loading of the sample covariance matrix is a popular technique applied to the minimum variance/minimum power distortionless response beamformer to increase robustness of the array system. However, this technique induces a trade-off between sidelobe suppression and the ability of the beamformer to adaptively cancel interference and reduce noise. Here we propose a new algorithm employing variable loading of the sample covariance matrix eigenvalues, which we show along with the standard diagonal loading technique to be a special case of a more general rational approximation problem. We also present an online implementation having computational complexity comparable to conventional methods, which in turn allows the weight vector to be efficiently updated. Simulation results indicate that in comparison with standard diagonal loading techniques, the proposed method exhibits enhanced robustness and performance.