二值无线传感网络下异常活动的分布式检测

针对无线传感网络下的异常活动检测问题,提出了异常活动分布式检测方法(distributed abnormal activity detection approach,DetectingAct). DetectingAct将活动的定义从轨迹扩展到轨迹和持续时间的组合,将异常活动定义为在数据分布上与正常活动,即数据中反复出现的活动,偏差较大的活动,利用节点自身计算资源和存储资源进行检测. DetectingAct采用时间相关的频繁项集挖掘算法(duration-dependent frequent pattern mining algorithm,DFPMA )从数据中挖掘正常活动. 算法采用了非监督学习方法,避免了监督学习需要大量标记数据的缺点;按分布式存储机制(distributed knowledge storage mechanism,DKSM)将正常活动模式存入各节点;用分布式检测算法(distributed abnormal activity detection algorithm,DAADA )检测活动. 理论分析和实验结果表明,分布式检测方法相比传统的活动检测算法,实时性更强,平均检测长度为轨迹的78.2%,精度更高,准确率达到96.9%. 

Distributed Abnormal Activity Detection in Binary WSNs

Distributed abnormal activity detection approach (DetectingAct), which employs the computing and storage resources of these sensor nodes, was proposed to detect abnormal activity under binary sensor network. In DetectingAct, activity was defined as the combination of trajectory and duration, while abnormal activity was defined as the activity whose deviation between normal activities, i.e. repetitive activities, is big enough. Firstly, DetectingAct found the normal activity patterns through duration-dependent frequent pattern mining algorithm (DFPMA), which adopted unsupervised learning instead of supervised learning. Secondly, the distributed knowledge storage mechanism (DKSM) was introduced to store the mined patterns in each node. Finally, Distributed abnormal activity detection algorithm (DAADA), which was based on the clustering analysis, was introduced to compare the present activity with normal activity patterns to determine the possibility of the current activity being abnormal. The feasibility, real-time property and accuracy of the approach were evaluated by experiments. The average detect distance reaches 78.2% and the accuracy is 96.9%.