Distributed state fusion estimation for nonlinear systems

Abstract

This paper investigates the distributed information fusion estimation problem for nonlinear systems. By using the classical extended Kalman filtering (EKF) and unscented Kalman filtering (UKF) methods, two distributed multi-sensor state fusion algorithms are presented for nonlinear systems in the information form. It is shown that the proposed extend information filter (EIF) based states fusion algorithm is equivalent to the centralized fusion algorithm in the information form. Finally, an example study of a target tracking system shows that the proposed distributed nonlinear fusion algorithm outperforms each local estimation, demonstrating the effectiveness of the proposed design methods.