Robust adaptive matched field processing

Matched field array processing is used in underwater acoustics situations with refraction and coherent multipath so that a plane-wave directional steering vector is inappropriate. In practice, the propagation conditions are seldom known with accuracy so that mismatch between the true signal and the model-based steering vectors is an important issue. A second issue is that the underlying dynamics of the environment make it impossible to obtain adequate sample support for covariance matrix estimation and inversion. To solve these problems, the robust dominant mode rejection algorithm is used with a white noise gain constraint for robustness. A closeness test to exclude an eigenvector and inhibit nulling when an eigenvector is too close to the steering vector is examined. The two-signal case is presented in detail. Relationships between the eigenvectors and the signal direction vectors are presented in terms of their relative strength and spatial correlation.