Encoding–decoding-based fusion estimation with censored measurements: When data transmission meets random bit errors

Abstract

This paper focuses on the state fusion estimation (FE) problem for a class of multi-sensor systems, where a specific measurement nonlinearity, namely censored measurements, is taken into account. The censoring phenomenon is described by the Tobit I model for practical engineering. Furthermore, in order to effectively alleviate the network communication burden and improve the reliability of data transmission, the binary encoding strategies (BESs) are employed in the communication channel from the sensors to the estimators. A sequence of Bernoulli random variables is used to model the random bit errors induced by channel noise during transmission. More specifically, an optimal fused state estimator is designed to integrate the benefits from multiple sensor outputs efficiently. In this paper, a FE scheme under BES is proposed to ensure that the FE error dynamics is exponentially bounded. Sufficient conditions for the existence of the desired local estimators and fusion estimator are firstly obtained, and then the optimal local estimator gains and the weighting matrices are acquired by solving certain optimization problems. Finally, the effectiveness of the estimation method is validated through a simulation example.