Sparsity based Space-Time Adaptive Processing using message passing

In this paper, we focus on the effective Space-Time Adaptive Processing (STAP) method in nonhomogeneous clutter environment. The nonhomogeneous clutter leads to the lack of sufficient training data for clutter covariance matrix estimation in traditional STAP methods. By utilizing the sparsity of the distribution of clutter in angle-Doppler domain, we build a factor graph model and develop a message passing algorithm to estimate the space-time distribution of clutter. The proposed method effectively reduces the number of training data compared with traditional methods. The numerical results show that the method outperforms the existing sparse recovery based STAP methods in nonhomogeneous clutter environment with higher accuracy and lower complexity.