Optimized Distributed Filtering for Time-Varying Saturated Stochastic Systems With Energy Harvesting Sensors Over Sensor Networks

This paper addresses the distributed filtering (DF) problem for time-varying saturated stochastic systems subject to energy harvesting (EH) sensors and time delay through sensor networks. The sufficient energy is a prerequisite for normal data transmission, so the EH technique is considered in the communication network, which can be regarded as an explicit decision, i.e., the sensors have the ability to harvest energy from surrounding environment. Particularly, the data information can be transmitted only when the sensors store nonzero units of energy, and vice versa. The specific probability distribution of EH level for individual sensor node can be computed iteratively at each sampling time by virtue of rigorous theoretical derivations. The focus is on the design of a novel DF scheme such that an optimized upper bound matrix on the filtering error covariance is obtained. Furthermore, the boundedness analysis with regard to the proposed filtering error dynamics is discussed with the help of some detailed mathematical computations. Finally, some comparative experiments are used to illustrate the validity of the developed variance-constrained optimized DF scheme under EH strategy.