High-resolution array processing using implicit eigenvector weighting techniques

Many high-resolution bearing estimators require the explicit calculation of the eigenvectors and eigenvalues of the cross-spectral matrix of the sensor outputs. Once the eigenvectors have been calculated, various estimators can be derived by altering the eigenvalues to give a reweighting of the eigenvectors. These weighting functions are reminiscent of ideal filter responses in analog filter theory, where practical filters are designed by using polynomial approximations to the ideal desired response. The approximation theory developed for filter design is used to derive high-resolution bearing estimators that do not require explicit calculation of the eigenvectors. >