On the divergence calculation of the Extended Kalman Filtering for chaotic signals

Abstract

The Extended Kalman Filter (EKF) has become one of the options to achieved synchronization of chaotic signals with important applications for wireless communications. However the joint operation of the EKF and the chaotic signals suffers from several reported drawbacks that negatively affect the performance of the EKF up to a breaking point. This paper shows a rather simple way of establishing a reference threshold that allows to avoid the breaking point and keeps the EKF in a working regime that yields good performance in terms of the Mean Square Error (MSE) between the real and estimated chaotic signals. Three different chaotic systems are used to illustrate the proposed approach.