A constraint projection approach for robust adaptive beamforming

When the optimal signal-to-interference-plus-noise-power ratio (SINR) at the output of the linearly constrained minimum variance beamformer is large, small perturbation errors in the steering constraint can cause substantial degradation of the SINR. Even if there is no perturbation, an excessive number of data snapshots can be required for the SINR to converge to its optimal value. To alleviate both the perturbation and sample data problems, a novel approach is proposed-projecting the nominal steering vector onto the signal-plus-interference subspace to obtain a new steering constraint. Analysis suggests that the problems of perturbation error and insufficient data are closely related