Joint Correlations Sparse Bayesian Learning STAP With Prior Knowledge of Clutter Ridge

Abstract

Space-time adaptive processing (STAP) based on sparse Bayesian learning (SBL) can significantly improve clutter suppression performance utilizing clutter sparsity. However, the existing SBL-STAP algorithms lack full use of correlations, which leads to unsatisfactory performance and slow convergence speed. In this article, we propose a joint correlations SBL-STAP (JCSBL-STAP) algorithm to improve clutter suppression performance. It comes from a rational idea that the clutter ridge in the space-time domain is not only the origin of clutter sparsity, but also the origin of correlations. Normally, the amplitude of scatterers along the clutter ridge are correlated between multiple samples and have clustered correlation properties in each sample. The JCSBL-STAP algorithm utilizes a joint correlations sparse prior to exploiting both correlations and provides a multisample correlation decoupling framework to update hyperparameters. The algorithm is executed on a proposed hybrid prior dictionary. Compared with the conventional uniform dictionary, the hybrid prior dictionary can easily express the clustered correlation properties and effectively alleviates the off-grid problem. Experimental results confirm the performance of the proposed method on both simulated data and measured Mountain-Top data.