基于分簇算法的无线传感网络异常时间序列数据过滤方法  

A Clustering Algorithm Based Method for Filtering Abnormal Time Series Data in Wireless Sensor Networks

作　　者：都美花 徐宝鸿 DU Meihua;XU Baohong(Gansu Vocational&Technical College Of Nonferrous Metallurgy,Jinchang 737100,China)

机构地区：[1]甘肃有色冶金职业技术学院,甘肃金昌737100

出　　处：《长江信息通信》2024年第7期65-67,共3页Changjiang Information & Communications

摘　　要：无线传感网络非常容易受到病毒攻击而产生异常时间序列数据,为了准确过滤网络中的异常时间序列数据,提出一种基于分簇算法的无线传感网络异常时间序列数据过滤方法。根据无线传感网络节点的能量消耗,利用分簇算法,将网络节点划分为多个簇头,引入网络节点的能量调节系数,确定网络节点周围的邻居节点数量,计算无线传感网络的最佳簇头。通过构建异常时间序列数据过滤的评判矩阵,给出异常时间序列数据的评判结果,根据无线传感网络的信息增益,选择异常时间序列数据的特征。根据无线传感网络的分簇情况,分类异常时间序列数据特征,通过计算异常时间序列数据的采集速度,得到异常时间序列数据的活动频率,实现异常时间序列数据的过滤。实验结果表明,文中方法能够过滤掉无线传感网络中的异常时间序列数据,并将过滤效率控制在90%以上。Wireless sensor networks are very susceptible to virus attacks and generate abnormal time series data.In order to accurately filter abnormal time series data in the network,a cluste-ring algorithm based method for filtering abnormal time series data in wireless sensor networks is proposed.Bascd on the cnergy consumption of wireless sensor network nodes,a clustering al-gorithm is uscd to divide the nctwork nodes into multiple cluster heads.The energy adjustment coefficient of the network nodes is introduced to determine the number of neighboring nodes a-round the network nodes,and the optimal cluster head of the wireless sensor network is calculat-ed.By constructing an evaluation matrix for filtering abnormal time series data,the evaluation results of abnormal time series data are presented.Based on the information gain of wireless sen-sor networks,the characteristics of abnormal time series data are selected.Based on the clustering situation of wireless sensor networks,classify the characteristics of abnormal time scries data,calculate the collection speed of abnormal time series data,obtain the activity frequency of ab-normal time series data,and achieve filtering of abnormal time series data.The experimental re-sults show that the proposed method can filter out abnormal time series data in wireless sensor networks and control the filtering efficiency to over 90%.

关 键 词：分簇算法 时间序列 异常数据 无线传感网络 过滤方法 计算簇头 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]