Reduced dimension principal components for STAP

This paper describes a new STAP algorithm based on Principal Components (PC) method. PC performs an eigendecomposition of the full space-time covariance matrix retaining the dominant (those associated with the interference space) eigenvalues and eigenvectors pairs to build the STAP filter coefficients. In this new proposed algorithm PC method is applied on the space-time covariance matrix associated with a subset of P, successive and overlapped, pulses of the full coherent train. The application of PC to the overlapped set of pulses inside the train is then followed by Doppler filtering. Simulations have been performed proving the effectiveness of the proposed algorithm showing near optimum SINR performances. The robustness of the new proposed algorithm has also been tested in presence of internal clutter motion (ICM) with again near optimum SINR performances. The new algorithm has demonstrated to further reduce, with respect to PC, the required sample support needed for the STAP filter together with less computational complexity.