Robust non-homogeneity detection algorithm based on prolate spheroidal wave functions for space-time adaptive processing

The estimated clutter covariance matrix is always corrupted by the interference target signals (outliers) in non-homogeneous clutter environments, which leads the performance of space-time adaptive processing (STAP) to be degraded significantly for clutter suppression. Therefore a robust non-homogeneity detection algorithm by utilising the prolate spheroidal wave functions (PSWF) is proposed to eliminate the outliers from the training samples set in this study, which can estimate the clutter covariance matrix more accurately for STAP. In the proposed method, the basis vectors of PSWF according to the system parameters are first calculated, which can be computed offline and stored in memory beforehand, and then the corresponding clutter covariance matrix is constructed. In the following, the constructed covariance matrix is combined with the generalised inner products (GIP) method to obtain the corresponding statistics. The training samples contaminated by the outliers are eliminated based on the comparison of the statistics and the designated threshold. By analysing the sensitive coefficients and the simulation results, it is found that the proposed method (PSWF-GIP) can more effectively eliminate the outliers and improve the performance of STAP in non-homogeneous clutter environments.