无线传感器网络中基于压缩感知和GM(1,1)的异常检测方案  

Abnormal Event Detection Scheme Based on Compressive Sensing and GM(1,1) in Wireless Sensor Networks

作　　者：李鹏[1] 王建新[1] 曹建农[1,2] 

机构地区：[1]中南大学信息科学与工程学院,长沙410083 [2]香港理工大学电子计算学系

出　　处：《电子与信息学报》2015年第7期1586-1590,共5页Journal of Electronics & Information Technology

基　　金：国家自然科学基金重点项目(61232001/F02);国家自然科学基金面上项目(61173169/F020802)资助课题

摘　　要：针对现有的异常事件检测算法准确率低和能量开销较大等问题,该文提出一种基于压缩感知(CS)和GM(1,1)的异常事件检测方案。首先,基于分簇的思想将传感器节点的数据进行压缩采样后传输至Sink,针对传感器网络中数据稀疏度未知的特点,提出一种基于步长自适应的块稀疏信号重构算法。然后,Sink基于GM(1,1)对节点发生的异常进行预测,并对节点的工作状态进行自适应调整。仿真实验结果表明,相比于其它异常检测算法,该算法的误警率和漏检率较低,在保证异常事件检测可靠性的同时,有效地节省了节点能量。In order to solve the problems of the low accuracy and the high energy cost by the existing abnormal event detection algorithm in Wireless Sensor Networks （WSN）, this paper proposes an abnormal event detection algorithm based on Compressive Sensing （CS） and Grey Model（1,1） （GM（1,1） ）. Firstly, the network is divided into the clusters, and the data ave sampled based on compressive sensing and are forwarded to the Sink. According to the characteristics of the unknown data sparsity in WSN, this paper proposes a block-spavse signal reconstruction algorithm based on the adaptive step. Then the abnormal event is predicted based on the GM（1,1） at the Sink node, and the work status of the node is adaptively adjusted. The simulation results show that, compared with the other anomaly detection algorithms, the proposed algorithm has lower probability of false detection and missed detection, and effectively saves the energy of nodes, with assurance the reliability of abnormal event detection at the same time.

关 键 词：无线传感器网络 异常事件检测 压缩感知 GREY Model(1 1)(GM(1 1)) 信号重构 能耗 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]