Distributed cubature Kalman filtering for nonlinear systems with stochastic communication protocol

Abstract This paper is concerned with the distributed cubature Kalman filtering (DCKF) for a class of discrete time‐varying nonlinear systems subject to stochastic communication protocol (SCP). In order to avoid the data collisions, the SCP is introduced to randomly schedule each sensor node information from one of the neighboring nodes to the filter which is presented via not only the information of itself sensor but also the information of neighboring sensors. The considered scheduling probability of the selected node is unknown. Therefore, the exact filtering error covariance is not available. The purpose of this paper is to design a DCKF under the spherical–radial cubature rule such that an upper bound of the filtering error covariance is guaranteed by utilizing fundamental inequality. Such an upper bound is dependent on known upper and lower bounds of the scheduling probabilities. Subsequently, the desired filter matrices are given by minimizing the upper bound of the filtering error. A sufficient condition is derived by using Riccati‐like difference equations method. Besides, a recursive form of filter algorithm is designed for online computation. Finally, the usefulness of the DCKF is verified by utilizing a simulation example on the induction machines.