Fault-tolerant sensor network based on fault evaluation matrix and compensation for intermittent observation

This paper deals with a fault-tolerant sensor network configuration problem for a target navigation. A sensor network system consists of many sensor nodes and its network connections. Each sensor node can exchange information by wireless communication. A disadvantage of this system's property is that if there is an inaccurate information from a faulty sensor, this information has possibilities to be diffused to other sensors. Therefore, it is important to reduce effects of the inaccurate information by detecting a faulty sensor as possible as we can. And moreover, in feedback control system for navigation, we have to consider a lack of control inputs which happens depending on each sensor's intermittent observation. We propose two estimation methods for constructing a fault tolerant system. Specifically, we propose a fault-evaluation matrix for the fault detection, and we define a novel switching rule for shutting off inaccurate measurement data. Then we also propose a compensation algorithm for the problem of intermittent observation by using an estimated observation value.