Distributed Kalman Filter Through Trace Proximity and Covariance Intersection
Xiaoxi Yan, Longhu Jin
Abstract
Distributed filtering algorithms have gained wide-spread application in wireless sensor networks attributed to their advantages of low communication overhead and strong robustness. We propose a novel distributed Kalman filter. The core idea is that each node selects an adjacent node with the minimum cost function, which is followed by data fusion between the two nodes through the covariance intersection method. The specific selection process significantly reduces the complexity of the algorithm, and the unique fusion method ensures data reliability while improving estimation accuracy. We demonstrate that the designed distributed Kalman filtering algorithm is unbiased and consistent. Simulation results are provided to verify the correctness and performance of our algorithm.