基于子图拉普拉斯谱的异常传感器检测及识别方法

Anomalous Sensor Detection and Recognition Based on Laplacian Spectrum of Sub-Graph

无线传感器网络中传感器异常检测是确保数据可靠性和系统正常运行的重要环节.将无线传感器网络用图模型描述,针对图上边缘区或稀疏区的异常传感器难以检测及识别的问题,本文提出了一种基于子图拉普拉斯谱的异常传感器检测及识别方法.该方法首先对系统图进行子图划分,再将图上信号转换至拉普拉斯谱信号,然后经低通滤波器处理,将图频域信号还原至节点域信号,通过比较还原信号与采集信号来判断子图的异常情况,最后对异常子图进行分析识别.基于公开数据集验证,本文所提方法对于无线传感器网络中单个异常传感器的检测率可以达到95%以上,其漏检率与误检率为15%以下,检测效果优于其他现有方法.

Detecting the existence of anomalous sensors in a wireless sensor network is important to ensure the reliability of the data and the normality of the system.Considering that it is difficult to detect and recognize the anomalous sensors in the edge or sparse area on the graph,which describes a wireless sensor network,this paper proposes a method for detecting and recognizing anomalous sensors based on Laplacian spectrum of sub-graph.The method first divides the system graph into several subgraphs,and then it converts the signals on the graph to Laplacian spectrum signals.This paper uses a low-pass filter to restore the graph signals in frequency domain to that in node domain,and then compares the restored signals with the acquired signals to judge the anomalous sub-graphs.Finally,the anomalous sub-graphs are identified by analyzing the data result.Through performance testing based on a public data set,for a single anomalous sensor in a wireless sensor network,the detection rate can reach more than 95%,the missed detection rate and false detection rate both are less than 15%,illustrating that the detection effect is better than other existing methods.