A waveform-agile unscented Kalman filter for radar target tracking

Abstract

This paper proposes a dynamic waveform selection algorithm for radar target tracking. Following the waveform auto-adaptive ideas in the classical control theory, the Cramer-Rao lower bound (CRLB) of the covariance of target range and range rate estimations is utilized to describe the statistic characteristics of the measurement noises in tracking. Then the relationship between waveform parameters and tracking performance is established. The CRLB of target estimations corresponding to a certain waveform is obtained through the ambiguity function. The unscented Kalman filter (UKF) is used as the tracker and a secondary UKF is used to predict the tracking MSE. Minimizing the tracking MSE is chosen as the criterion of the dynamic waveform selection. At every time step of tracking, optimal transmitted waveform parameters are selected to track the nonlinear 2D target. Simulation results show the algorithm can improve the tracking performance when target states and measurements are nonlinear.