摘要： 针对用于事件检测的带状无线传感器网络,间歇性故障检测易误判事件为故障,以及在高故障率区域误判率较高的问题,提出一种适合带状传感器网络的动态分布式故障检测算法。该算法基于分簇和决策扩散策略,在簇内找到正常的簇头,以此为参照判断簇成员的状态。利用 间歇性故障和事件在时间域的特性,排除一部分故障节点成为簇头的可能性,使用空间相关性和动态时间弯曲的方法鉴别事件区域的正常节点和故障节点,从而筛选出正常的簇头。实验结果证明,在高故障率区域和事件区域,该算法能达到99%以上的检测精确度,5.5%以下的误 判率,并且具有较低的能耗水平。

关键词: 带状无线传感器网络, 间歇性故障检测, 时空相关性, 事件检测, 动态时间弯曲

Abstract: Intermittent fault detection tends to diagnose events as faults in strip-based wireless sensor network applied to event detection,and also to decrease the higher false alarm rate in high fault probability region,a Dynamic Distributed Fault Detection(DDFD) algorithm is proposed for wireless strip-based sensor network.This algorithm is based on cluster and decision diffused schemes by finding normal cluster head and using it as reference to diagnose the statuses of cluster members.A part of the fault sensors are excluded to become the cluster head based on the features of intermittent fault and events in time domain.Faults and events are distinguished by spatial correlation and Dynamic Time Warping(DTW).Thus normal cluster head is selected.Experimental results prove that this algorithm can achieve over 99% detection accuracy,below 5.5% false alarm rate,and has lower energy consumption in the case of higher sensor fault probability or events.

Key words: strip-based wireless sensor network, intermittent fault detection, temporal and spatial correlation, event detection, Dynamic Time Warping(DTW)

中图分类号: 

• TP393