Distributed robust filtering with hybrid consensus strategy for sensor networks

Abstract

The problem of distributed state estimation for time-varying uncertain systems over a sensor network within the robust Kalman filtering framework is studied. It is assumed that the parameters of the underlying model are subjected to stochastic uncertainties. At the first step, the authors present a particular form of the Kalman filter named information form of a robust Kalman filter. Then, using a consensus scheme that guarantees an agreement in estimation among nodes, they propose a new consensus-based robust filtering approach. The consensus methodology is a hybrid method that is a combination of consensus on information (CI) and consensus on measurement (CM). To this end, before proposing the hybrid consensus filtering algorithm, distributed robust filtering based on CI and CM have been presented. Next, they show that the estimation error in the proposed consensus algorithm is exponentially mean square bounded using a Lyapunov function. Finally, they provide two illustrative examples to show the robust performance and effectiveness of the proposed consensus filtering algorithms.