摘要
传感器网络的异常数据检测对于相关应用具有非常重要的意义。针对异常数据问题,提出了一种基于数据变化模式和数据空间相关性的无线传感器网络分布式异常数据检测方法。该方法首先为数据创建一个分布模型,即将数据变化模式映射到某个特征空间的分区中,确定非频繁分区,再通过判定数据是否落在非频繁分区来筛选出潜在异常值,最后根据邻居节点数据的空间相关性对异常值产生的原因进行判定。实验结果表明,该方法在避免较高运算复杂度的同时,能够有效地检测并区分网络中的错误数据与异常事件。
Sensor networks anomaly data detection has a very important significance for the relevant application. Aiming at the problem of ab- normal data, this paper proposes a distributed anomaly data detection method for wireless sensor networks based on data change pattern and da- ta spatial correlation. The method first creates a distribution model for the data, that is, the data change pattern is mapped to the partition of a feature space, determines the infrequent partition, and then determines whether the data falls on the infrequent partition to filter out the poten- tial anomaly value. Finally, according to the spatial correlation of data between neighborhood nodes to determine the cause of the outliers. The experimental results show that the proposed method can effectively detect and distinguish the wrong data and abnormal events in the network while avoiding the high computational complexity.
出处
《信息技术与网络安全》
2018年第1期70-75,共6页
Information Technology and Network Security
基金
国家自然科学基金(61370210)
关键词
无线传感器网络
异常检测
分布式策略
数据变化模式
空间相关性
wireless sensor networks
anomaly detection
distributed strategy
data change pattern
spatial correlation
