DOA Estimation by Off-Grid Compressive Sampling Matching Pursuit with Impulsive Noise

A sparse based DOA mathematical model is proposed based on the vectorization of a fractional low-order covariance (FLOC) matrix. The sparse recovery-based method of off-grid compressive sampling matching pursuit was adopted to perform the direction-of-arrival (DOA) estimation of the target. In this study, the FLOC matrix was vectorized to construct an over-complete dictionary. The CoSaMP algorithm was improved and successfully applied to DOA estimation. In addition, the alternating iteration method is applied to solve the off-grid model. The proposed algorithm can achieve high-accuracy DOA estimation at low generalized signal-to-noise ratio (GSNR) and small snapshots. The simulation experiments results demonstrated the effectiveness and estimation accuracy improvement of the proposed algorithm in impulsive noise.