Sensor Fault Detection Algorithm Based on Credibility and Neighbor-Cooperation

LIN Yan-fei,SHAO Su-jie,ZHONG Cheng,GUO Shao-yong,QIU Xue-song

DOI: https://doi.org/10.11959/j.issn.1000-436x.2016170

2016-01-01

Abstract: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.