A novel robust minimum error entropy Kalman filter in the presence of measurement packet dropping

Abstract

In a practical system, the simultaneous presence of packet dropping and impulsive noise can render the state estimation problem complicated and its solution challenging. Kalman-type filter algorithms based on the minimum error entropy may involve instabilities in numerical calculations. To efficiently process the impulse noise in the system, this paper proposes a novel robust minimum error entropy Kalman filter algorithm in the presence of packet dropping, named RMEEPDKF. The augmented model of the packet dropping system is constructed by designing a coefficient matrix containing the packet dropping variable. In addition, the posterior state estimation is updated based on the fixed-point iterative algorithm with the robust minimum error entropy criterion. Furthermore, we extend RMEEPDKF to the suboptimal Kalman filter with packet dropping to obtain a estimator named RMEEPDSKF. The simulation results demonstrate that RMEEPDKF and RMEEPDSKF can achieve the highest estimation accuracy when packet dropping and impulse noise coexist in the system, which implies that the introduction of a small amount of computational burden may be acceptable to achieve a high estimation accuracy.