无线传感器节点的故障诊断研究  

Research on Fault Diagnosis of Wireless Sensor Nodes

作　　者：尚兴宏[1,2] 尚曦乐 章静 钱焕延[1] 

机构地区：[1]南京理工大学计算机科学与技术学院,南京210094 [2]上海华为技术有限公司,上海201206 [3]上海金洲小学,上海200333

出　　处：《计算机科学》2013年第06A期327-329,343,共4页Computer Science

基　　金：国防科工委应用基础资金资助

摘　　要：无线传感器网络节点发生故障不仅消耗节点的能量和网络带宽,甚至会造成网络瘫痪。在分析无线传感器网络节点故障类别的基础上,分别使用相关向量机、支持向量机等算法对其进行研究,并用节点的特征值及相应的故障类型训练相关向量机及支持向量机的分类器。仿真结果表明,相关向量机比支持向量机和人工神经网络有更高的诊断精度。In Wireless Sensor Networks（WSNs） ,faulty sensor nodes may consume the limited energy and bandwidth of network. Furthermore, they will produce the paralysis of entire systems. In this paper, RVM and SVM algorithm was applied to fault diagnosis for sensor nodes based on the analysis of fault type of WSNs nodes, and then the values of the features and the corresponding fault types of wireless sensor nodes were used to train RVM and SVM classifier. Simulation results show that the diagnosis results of the RVM model for wireless sensor are better than those of SVM and ANN.

关 键 词：无线传感器网络 故障诊断 相关向量机 支持向量机 人工神经网络 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]