Global state estimation under sequential measurement fusion for clustered sensor networks with cross-correlated measurement noises

In practical multi-sensor fusion estimation, both the expensive communication burden of long-distance connection sensor networks and cross-correlated noises are usually inevitable, and need to be taken into account. To adequately address these difficulties, a global fusion estimation problem is studied for clustered sensor networks with cross-correlated measurement noises. To begin, we derive direct and indirect sequential measurement fusion (SMF) schemes as fusion rules, relying on whether a weighted least square method is used directly for the arriving measurements. We discover that the direct SMF has lower computational complexity than the indirect SMF, but the estimation accuracy of the latter is higher when the number of sensors exceeds three. Then, a globally indirect sequential measurement fusion estimation (GISMF) algorithm is proposed by using the indirect SMF, which is more computationally efficient and suitable for asynchronous fusion than that of batch fusion estimation. Accuracy relations among the GISMF, the batch state fusion estimation, and the centralized Kalman fuser (CKF) are proved. Moreover, the equality of the estimation accuracy between the GISMF and the CKF is also performed if all the measurement noises are mutually independent or if the measurement noises are only cross-correlated within the same cluster. Finally, several simulation results are implemented to show the effectiveness of the presented algorithm. (C) 2022 Elsevier Ltd. All rights reserved.

Automated summary

This paper investigates the global fusion estimation problem in clustered sensor networks with cross-correlated measurement noises. It proposes a globally indirect sequential measurement fusion estimation algorithm and proves the accuracy relations with batch fusion estimation and centralized Kalman fuser.