论文部分内容阅读
To reduce excessive computing and communication loads of traditional fault detection methods,a neighbor-data analysis based node fault detection method is proposed.First,historical data is analyzed to confirm the confidence level of sensor nodes.Then a node’s reading data is compared with neighbor nodes’ which are of good confidence level.Decision can be made whether this node is a failure or not.Simulation shows this method has good effect on fault detection accuracy and transient fault tolerance,and never transfers communication and computing overloading to sensor nodes.
To reduce excessive computing and communication loads of traditional fault detection methods, a neighbor-data analysis based node fault detection method is proposed. First, historical data is analyzed to confirm the confidence level of sensor nodes. Shin a node’s reading data is compared with neighbor nodes’ which are of good confidence level .Decision can be made whether this node is a failure or not.Simulation shows that this method has good effect on fault detection accuracy and transient fault tolerance, and never transfers communication and computing overloading to sensor nodes.