基于时间序列相似度的无线传感网故障检测算法

针对无线传感网中节点故障率高而导致检测精度低以及能耗大的问题,提出了基于时间序列相似度的无线传感网故障检测算法(FDABTS2).该算法利用节点时间序列进行差值和相似度的计算,剔除一定故障类型的节点,并对瞬时读数故障数据进行修正.当节点为可疑节点时,利用空间相关性,与剩余邻居节点协作完成判定.在仿真环境下,与LEFD、NADST以及传统的DFD故障检测算法相比,检测精度分别提高了13％、17％和25％,误报率分别降低了8.4％、21.4％和25.4％,网络能耗分别降低了2.9％、8％和21.3％,即使在瞬时读数节点故障率高达60％时,误报率也能分别降低1.9％、26.2％和38.2％.实验结果表明,该算法在保证高检测精度的情况下,仍能有效降低能耗,延长网络生命周期.

Fault detection algorithm based on time series similarity in WSN

Aiming at the problem of low detection accuracy and high energy consumption caused by the high node failure rate in the wireless sensor network, this paper proposed a failure detection algorithm based on time series similarity(FDABTS2). The algorithm used the time series of nodes to calculate the difference and similarity, eliminated nodes with a certain failure type, and corrected the instantaneous reading failure data. When the node was a suspicious node, it used the spatial correlation to determine with the remaining neighbor nodes. In the simulation environment, compared with LEFD, NADST and tradition fault detection of algorithm DFD, the node detection accuracy was improved by 13%, 17% and 25%, and node false alarm rate was reduced by 8.4%, 21.4% and 25.4%, respectively. The network energy consumption was reduced by 2.9%, 8% and 21.3%, respectively. Even when the instantaneous reading node failure rate was as high as 60%, the false alarm rate could be reduced by 1.9%, 26.2% and 38.2%, respectively. Experimental results show that the algorithm can effectively reduce energy consumption and extend the network life cycle while ensuring high detection accuracy.