Weighted knowledge-aided space time adaptive processing in heterogeneous environments

Since training samples are not always identically distributed with the clutter in the cell under test (CUT) in heterogeneous clutter environments, which lead to an inaccurate estimate of clutter covariance matrix (CCM), the performance of space time adaptive processing (STAP) degrade significantly. In order to improve the performance of STAP, we propose a robust weighted knowledge-aided (KA) STAP algorithm. Unlike the same weights used for sample covariance matrix estimation, the presented algorithm uses different weights on training samples for CCM, where the weights are developed based on the similarities between the a priori covariance matrix of the CUT and that of the training data. The algorithm is capable of measuring the effect of heterogeneous samples on CCM more properly, and then we obtain a covariance matrix which reflects the clutter property of CUT more accurately. As a result, the performance of weighted KASTAP is better than conventional STAP in clutter suppression. Experimental results using the real data verify the effectiveness of the proposed method.