Event-triggered information fusion for networked systems with missing measurements and correlated noises

This paper is concerned with the state estimation fusion problem for networked systems subject to bandwidth constraints and network-induced measurement missing. The stochastic networked system with correlated noises is firstly transformed to an equivalent one, whose process noise is independent of sensor measurement noises. To increase the efficiency of limited communication resource, a closed-loop event-triggered mechanism is designed based on the measurement innovation and Mahalanobis transformation, where data transmission is triggered only when the predefined condition is satisfied. The phenomenon of measurement missing is characterized as a series of independent Bernoulli sequences. For every sensor subsystem, optimal local event-triggered state estimators including filter, predictor and smoother are given in terms of Minimum Mean Square Error (MMSE). The statistical knowledge of event-triggered sampling and missing measurements is utilized in the design of local estimators. Then, based on the linear minimum variance criterion, the corresponding globally optimal event-triggered fusion estimators are proposed by a matrix-weighted combination of all available local estimates from sensor subsystems. The proposed decentralized fusion estimators have good abilities in reliability, fault tolerance and computation efficiency due to the netted parallel structure. The effectiveness and feasibility of the proposed algorithms is demonstrated by a numerical example.