M-Estimation-Based Robust Kalman Filter Algorithm for Three-Dimensional AoA Target Tracking

Abstract

An attractive issue in numerous applications is to track targets in three-dimensional (3-D) space with angle of arrival (AoA) measurements through nonlinear filters. The tracking performance degradation caused by outlier prompts a variety of robust filters. In this paper, an M-estimation-based robust bias compensation Kalman filter algorithm (MR-BCKF) is proposed. This algorithm recasts the AoA measurement equation to a linear form by pseudo-linearization, and then incorporates the M-estimation criterion into pseudo linear Kalman filter to enhance robustness, followed by the bias compensation to improve tracking accuracy. An improved three-segment weight function based on Mahalanobis distance is established to handle outliers for each element, which does not require the noise characteristics. Simulation demonstrates that MR-BCKF has enhanced robustness against outliers at different levels and achieves more accurate tracking compared with other robust filters.