Fast Converging and Controllable Structure- Aware Clutter Suppression Method for Airborne Polarimetric Array Radar

Abstract

Polarimetric space-time adaptive processing (PSTAP) significantly enhances the ability to detect low-speed targets for airborne early warning radar. However, incorporating diverse polarization sensor data poses challenges: it expands the snapshot dimension and increases clutter heterogeneity. Therefore, the performance of PSTAP may suffer due to the inaccurate estimation of the clutter-plus-noise covariance matrix (CNCM) with finite samples. Leveraging the Kronecker product structure of clutter, statistical framework-based Kronecker estimators can reduce sample requirements while maintaining the clutter suppression performance. But for low-speed target detection, we theoretically illustrate the limitations of this type of estimator. While sparse recovery (SR) space-time adaptive processing (STAP) methods can achieve satisfactory CNCM estimation with very few samples, they cannot be directly applied to PSTAP. In this article, we propose a structure-aware (SAW) sparse Bayesian learning (SBL) algorithm for PSTAP, named SAW-SBL PSTAP. By exploiting the independence between the polarization domain and the space-time domain, along with the intrinsic sparsity of clutter in the angle-Doppler plane, we model a block SR problem and develop a fast and controllable learning framework. This framework alternately updates the noise power, polarization covariance matrix, and clutter space-time power, resulting in precise CNCM estimation. Both simulated and measured data experiments verify the effectiveness and robustness of the proposed method, particularly in enhancing detection performance for low-speed targets.