Robustness of adaptive array processing

Summary form only given, as follows. Optimal or adaptive array processing has several advantages for applications to acoustic array processing in the ocean. These advantages over conventional methods include better sidelobe control and better noise rejection. The disadvantages include sensitivity to modeling errors directly proportional to the maximum array output signal-to-noise ratio (SNR), and the requirement to know the cross spectral covariance matrix which must be estimated from the data for any real-world applications in the ocean. The authors investigate the robustness of the minimum variance distortionless constraint (MVDC) estimator and so-called robust variants to this algorithm to modeling errors and errors in estimation of the covariance matrix. In particular, they study the sidelobe and nulling behavior and resulting array output SNR versus various levels of these errors.<>