Sequential Inverse Covariance Intersection Fusion Kalman Filter for Multi-sensor Systems with Packet Dropouts and Multiplicative Noise

Abstract

This paper concerned with the hybrid network system with state multiplicative noises, packet dropouts and linear correlation additive white noises. Firstly, the dimension of the original system is augmented by the method of augmented matrix. By introducing the fictitious noises to compensate the uncertainty of multiplicative noises and transform the augmented system into a standard system with only determined noise variance. Then, based on the standard system, local steady-state Kalman estimators are designed by using Lyapunov equation approach, and the SICI steady-state Kalman fusers (predictor and filter) are presented by using the sequential inverse covariance intersection algorithm. In this paper, two SICI steady-state Kalman fusers are presented combined with a hybrid networked system, which expand the application fields of the new ICI information fusion algorithm, which overcomes the shortcomings that the large amount of calculation given by the original CI fusion algorithm, and overcomes the large conservation of the original SCI fusion algorithm. It is proved that the accuracy of SICI fuser is higher than that of SCI fuser. Finally, some simulation results are given to verify the effectiveness of the proposed algorithm.