Measurement: Sensors (Dec 2021)
A Bayesian approach to distributed optimal filtering over a ring network
Abstract
This paper is concerned with the state estimation over a sensor network. Distributed estimation algorithms enable us to estimate the system state using the information from other sensors, even when the state is not completely observable from some sensors. The extension of the Kalman filter to the distributed case has been actively studied for the last decade. Many of the previous distributed Kalman filtering algorithms were proposed by incorporating a consensus control in the observer structure. However, these methods do not guarantee the optimality of the state estimates, and it turns out that their estimation accuracy will be significantly deteriorated even for a very simple network topology. To overcome this difficulty, we formulate the distributed state estimation problem based on the Bayesian inference. Then, we derive the optimal distributed estimation algorithm for a sensor network with the ring topology. The effectiveness of the proposed algorithm is also investigated by numerical experiments.