LAUSR

Abstract An improved THREE-Stage Extended Kalman Filter (3S-EKF) is presented in this paper for slow target tracking by collecting azimuth and elevation angle measurements on a single platform. The 3S-EKF is derived by measurement decouple and accurate prediction of relative height. Since the one-stage extended Kalman filter has measurement coupling issues, a TWO-Stage Extended Kalman Filter (2S-EKF) is proposed. It first estimates the 2D-projected target position using the azimuth angle only followed by updating relative height using elevation measurements. However, 2S-EKF ignores the geometrical relationship between the 2D-projected target position and relative altitude, making the prediction of relative height inaccurate. Thus, the 3S-EKF algorithm is proposed in this paper based on the exploration of the geometrical constraint to improve the prediction accuracy of relative height. The proposed 3S-EKF algorithm can achieve the Bayesian Cramer-Rao lower bound. Furthermore, the theoretical bias is analyzed with various target speeds. Simulation results demonstrate that the 3S-EKF algorithm has better estimation accuracy.