Distributed Information Filtering for Wireless Positioning in Correlated Measurement Noises

摘要

In this paper, distributed information filtering architectures for wireless sensor networks or location systems in correlated measurement noises are presented. To process sensor data in uncorrelated measurement noises, information filters can effectively integrate location estimates from multiple sensor groups. However, for correlated measurement noises, of which the covariance matrix is no longer in diagonal form, the performance of information filtering may degrade and become unacceptable. By using Cholesky factorization process, we reformulate the observation model as an equivalent pseudo form with uncorrelated noises, and apply the derived model to form an equivalently distributed information filtering architecture. Based on the derived Cholesky decomposition form, a hybrid time-difference of arrival / angle of arrival (TDOA/AOA) wireless location system is used in investigating the performance of distributed information processing. Simulation results show that the proposed architecture effectively achieves improved location and tracking accuracy.