Distributed Moving-Horizon Estimation with Event-Triggered Communication over Sensor Networks

This paper is concerned with the distributed moving-horizon estimation problem for linear discrete-time systems over wireless sensor networks. To reduce the communication burden, an event-triggered communication mechanism is proposed to govern the transmission of one-step predictions and measurements in each node. Under such a mechanism, a novel event-triggered distributed moving-horizon estimator is designed to deal with the state estimation problem subject to noise and state constraints. Each node computes its local state estimate by minimizing a cost function, which includes the event-triggered one-step predictions and measurements from the neighbors. It is proved that the developed distributed estimator is stable with the uniformly bounded estimation error in each node, provided by network connectivity, regional observability and small enough triggering parameters. A numerical simulation is given to verify the effectiveness of the proposed estimation algorithm.