The extended Kalman filter (EKF) has been widely used as anonlinear filtering method for radar tracking problems. However, it hasbeen found that in case cross-range measurement errors of the targetposition are large, the performance of the conventional EKF degradesconsiderably due to non-negligible nonlinear effects. In this paper, anew filtering algorithm for improving the radar tracking performance isdeveloped based on the fact that the correct evaluation of themeasurement error covariance can be made possible by doing it withrespect to the Cartesian state vector. The resulting filter may beviewed as a modification of the EKF in which the variance of the rangemeasurement errors is re-evaluated at each time step and themeasurements are sequentially processed in the order of azimuth andrange. Computer simulation results show that the proposed methodachieves superior performance than other existing filters whilerequiring a relatively small computational load
展开▼
