Performance Comparison of AKF and EFRLS in Tracking Dynamic Targets
Jie Zhou · Communications technology · 2009
Kalman filtering is the optimal state estimation with recursive form,and the optimal performance of Kalman filtering could be obtained only by knowing the information of the process noise and measurement noise. However, in many practial problems for tracking dynamic targets the noise is uncertain. Thus, it is necessary to develop new algorithms and adapt the new problems. This paper describes the estimation of process noise covariance by adaptive Kalman filtering(AKF) and the estimation of the unknown state by recursive formula of the extended forgetting factor recursive least squares(EFRLS). Through some simulations, the performances of these two methods in tracking dynamic targets are compared under the criterion of the minimum average error. The simulation results show that EFRls could satisfactorily track the dynamic targets in condition of correlated noises, and this adaptive Kalman filtering method could aquire better performance than EFRLS in condition of uncertain noises.