通过可疑节点与邻居节点的历史数据比较，判断节点是否故障是传感器故障检测的有效方法之一。该方法的关键点在于可疑节点的判断问题以及故障诊断消息发送时机的决策问题。为此，提出一种基于可信度和邻居协作的传感器故障检测算法。首先建立可信度模型进行节点可疑状态判定；随后基于邻居协作的思想向邻居节点发送故障诊断请求，通过等概率时间窗口决定该可疑节点诊断请求的发送时机；最后依据邻居节点的诊断回复进行故障情况的分类以及状态判断。通过仿真验证，该算法能够在降低故障诊断次数和数据拥塞率的前提下，提高故障检出率。%To effectively detect whether a node was failure, one of the effective sensor fault detection methods was to compare historical data of suspicious node with its neighbors. The key points of this method were identification of suspi-cious nodes, as well as decision problem of sending timing for fault diagnosis messages. To this end, a sensor fault detec-tion algorithm was presented based on credibility and neighbor-cooperation. Firstly, a credibility model to determine whether nodes are suspicious was established. Then suspicious nodes send fault diagnosis requests to neighbor nodes based on neighbor-cooperation. The sending timing was determined by equal probability time window. Finally, it could finish classification of fault condition and state judgment based on diagnosis responds of neighbor nodes. Simulation ex-periments show that this algorithm can achieve higher fault detection rate with less fault diagnosis times and low data congestion probability.
展开▼
