An Improved Optimal Subset Reduced-rank Method

An improved reduced-rank approach was presented to cope with the defections of high computational complexity, lack of training samples and clutter heterogeneity, existing in space-time optimum processor. Firstly, spatial spectral correlation coefficient was employed to depict the separation between object and clutter trajectory, and thus the selection of antenna-pulse pairs were converted to the minimization of the spatial spectral correlation coefficient. This step reduced the computational complexity and impression of clutter heterogeneity. Secondly, eigenvalue decomposition was implemented on the clutter-plus-noise covariance matrix, and eigenvectors were selected based on cross spectral to consist reduced-rank matrix. This step reduced the need for training samples. Finally, simulations prove the validity of proposed method.