Robust STAP Based on Training Snapshot Selection and Steering Vector Estimation

Abstract

The presence of mismatch between the presumed and actual target direction in spatial-temporal plane and the contamination of signal of interest (SOI) in training snapshots significantly degrades the performance of space-time adaptive processing (STAP). To solve the aforementioned problems, a robust STAP by employing the training snapshot selection and target direction correction is proposed in this paper. To solve the inaccurate estimation of clutter covariance matrix (CCM), training snapshots contaminated by target are eliminated using generalized inner products (GIP) statistics, and the more accurate CCM is obtained using the updated training snapshots. Then, an accurate steering vector (SV) estimation method is introduced using the prior knowledge of SOI. Finally, we obtain the proposed robust STAP weight with the novel CCM and the corrected target space-time SV. We carried out several computer simulations to validate the performance of our proposed method.