Fusion Predictors for Multisensor Stochastic Uncertain Systems With Missing Measurements and Unknown Measurement Disturbances

This paper addresses the information fusion state estimation problem for multisensor stochastic uncertain systems with missing measurements and unknown measurement disturbances. The missing measurements of sensors are described by Bernoulli distributed random variables. Measurements of sensors are subject to external disturbances whose any prior information is unknown. Stochastic parameter uncertainties of systems are depicted by multiplicative noises. For such complex systems with multiple sensors, the Kalman-like centralized fusion and distributed fusion state one-step predictors (i.e., prior filters) independent of unknown measurement disturbances are designed based on the linear unbiased minimum variance criterion, respectively. Estimation error cross-covariance matrices between any two local predictors are derived. Their steady-state properties are analyzed. The sufficient conditions for the existence of the steady-state predictors are given. Two simulation examples show the effectiveness of the proposed algorithms.