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.
出处
《计算机科学》
CSCD
北大核心
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
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
Wireless sensor networks(WSNs), Fault diagnosis, Relevance vector machine(RVM), Support vector maehine(SVM) ,Artificial neural networks(A_NN)