作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程

• 移动互联与通信技术 • 上一篇    下一篇

基于K-means聚类的WSN异常数据检测算法

费欢a,b,李光辉a,b   

  1. (浙江农林大学 a.信息工程学院; b.浙江省林业智能检测与信息技术研究重点实验室,浙江 临安 311300)
  • 收稿日期:2014-07-14 出版日期:2015-07-15 发布日期:2015-07-15
  • 作者简介:费欢(1990-),男,硕士研究生,主研方向:无线传感器网络;李光辉(通讯作者),教授、博士。
  • 基金资助:
    国家自然科学基金资助项目(61174023);浙江省自然科学基金资助项目(Y1110791)。

Abnormal Data Detection Algorithm for WSN Based on K-means Clustering

FEI Huan a,b,LI Guanghui a,b   

  1. (a.School of Information Engineering;b.Zhejiang Provincial Key Laboratory of Intelligent Monitoring in Forestry and Information Technology,Zhejiang A& F University,Lin’an 311300,China)
  • Received:2014-07-14 Online:2015-07-15 Published:2015-07-15

摘要: 为提高无线传感器网络应用系统的可靠性,对传感器节点采集的环境数据集进行检测,提出一种改进的异常数据检测算法。采用K-means算法思想,结合无线传感器网络数据的特点,以欧式距离作为指标,比较数据点的相似度并划分聚类,根据数据点与聚类中心之间的距离区分正常数据与异常数据。实验结果表明,当数据规模超过1 000时,与基于噪声的密度聚类算法相比,该算法对于异常数据的检测率较高,误报率较低。

关键词: K-means算法, 无线传感器网络, 聚类, 异常数据检测, 密度聚类

Abstract: In order to improve the reliability of Wireless Sensor Network(WSN) application system,it detects abnormal data from sensor environmental data set.An algorithm of abnormal data detection based on clustering of data mining is proposed in the paper,which not only adopts K-means clustering but also takes the characteristics of WSN data into account.This algorithm uses Euclidean distance to compare similarity of data for cluster partitioning,and identifies the abnormal data according to the distance between data point and cluster center.Experimental results show that when data is more than 1 000,compared with the algorithm based on Density-based Spatial Clustering of Applications with Noise(DBSCAN),the detection accuracy of this algorithm is higher and the false positive rate is lower under the same conditions.

Key words: K-means algorithm, Wireless Sensor Network(WSN), clustering, abnormal data detection, density clustering

中图分类号: