Abstract:

The performance of target detection decreases in space-time adaptive processing (STAP) when inhomogeneity of clutter power is produced with training samples contaminated by targetliking signals. In allusion to this problem, an inhomogeneous sample selection method for STAP based on matrix similarity is proposed. Firstly, mean Hausdorff distance is adopted to measure the similarity of the sample matrix in this method through the difference measurement of contaminated training samples and uncontaminated training samples. Then the similarities of different samples are calculated with the convex optimization package. The contaminated samples are rejected on the basis of different similarities. The simulation result shows that the proposed method is more effective for contaminated samples detection because of small power targetliking signals compared with the generalized inner product (GIP), which eliminates the influence of small power targetliking signals on the covariance matrix estimation and improves the target detection performance for STAP in inhomogeneity of clutter power environment.