基于DBSCAN的环境传感器网络异常数据检测方法

ABNORMAL DATA DETECTION METHOD FOR ENVIRONMENT WIRELESS SENSOR NETWORKS BASED ON DBSCAN

随着传感器网络环境监控应用的发展,传感器网络测量数据的异常检测近年来受到学术界和工业界的高度关注。提出一种基于DBSCAN(Density-Based Spatial Clustering of Application with Noise)的异常数据检测方法,该方法利用距离定义数据的相似度进行划分聚类,使用DBSCAN算法提取环境特征集,并根据特征集对异常数据进行检测。最后,基于真实的传感器网络完成了多组实验,实验结果表明该方法能够实时准确地检测出异常数据。

With the development of applying sensor network to environment monitoring, the abnormal detection on data measurement in sensor network attracts much attentions recently by both academics and industry. A method of abnormal data detection based on DBSCAN （Density-based spatial clustering of application with noise） is proposed in the paper, which uses distance to define the similarity of data for cluster partitioning, and uses DBSCAN to extract the feature set of environment, and to detect the abnormal data according to the feature set. In the end of the paper we present a set of experiments accomplished in real sensor network, the experimental results show that the proposed niethod can detect the abnormal data timely and correctly.