Iteratively reweighted lp norm minimization for DOD and DOA estimation in bistatic MIMO radar under impulsive noise

In this paper, an iterative-reweighting-based algorithm is proposed to improve the performance of direction of-departure (DOD) and direction-of-arrival (DOA) estimation in bistatic MIMO radar under impulsive noise. The BSUM-RELAX algorithm is generalized to its two dimension and multiple measurement vectors (MMV) version for robust DOD and DOA estimation. The optimization problem is cast to an lp (with 0 < p < 2) norm minimization problem to alleviate the impact of the impulsive noise, which is iteratively solved by minimizing the objective function based on each variable. Nevertheless, such a problem is of non-convexity and highly depends on a good guess of all the initial values of DODs and DOAs of multiple targets in order to obtain considerable performance. Therefore, an off-grid two-dimensional Newtonized orthogonal matching pursuit (2D-NOMP) algorithm is developed to obtain a near-optimal initial guess. Simulations are conducted to verify the robustness of the proposed approach for DOD and DOA estimation under impulsive noise.