Robust space-time adaptive processing based on covariance matrix reconstruction and steering vector correction

Abstract

Clutter presents considerable heterogeneity in forward-looking airborne radar (FLAR) applications and conventional space-time adaptive processing (STAP) methods are sensitive to model mismatch. As a result, when a strong target signal contaminates the training samples, despite the use of guard cells, the performance of conventional STAP methods degrades significantly. In this study, a robust method, which involves reconstructing a target-free covariance matrix and correcting the presumed steering vector to prevent target cancellation in FLAR, is proposed. First, the target-free covariance matrix is reconstructed through integrating the spatial–temporal spectrum over a sector separated from the desired frequency and direction of targets. Subsequently, the mismatch between presumed steering vector and actual steering vector is corrected via quadratic optimisation. In addition, the processing scheme is applied to real-measured clutter data, and the experimental results validate the effectiveness of the proposed method.

Keywords

• covariance matrices
• space-time adaptive processing
• array signal processing
• radar signal processing
• airborne radar
• radar clutter
• radar detection
• conventional space-time adaptive processing methods
• strong target signal
• conventional stap methods degrades
• robust method
• target-free covariance matrix
• presumed steering vector
• target cancellation
• flar
• actual steering vector
• processing scheme
• robust space-time adaptive processing
• covariance matrix reconstruction
• steering vector correction
• considerable heterogeneity
• airborne radar applications