事件监测是无线传感器网络的一种重要应用。针对该应用中软故障节点提供的错误数据会降低监测的准确性的问题,提出了一种分布式的容错事件边界检测算法。节点只需与邻节点交换一次传感数据,通过简单地计算识别故障;正常的事件节点利用统计比较的方法判断其是否处于事件边界,边界宽度可根据网络用户的要求调节。该算法执行时所需的通信量小,计算复杂度低,时延小,对大规模网络具有很好的可扩展性。仿真结果表明即使节点故障率很高,应用该算法仍可以获得很好的检测效果。
Event detection is an important application of wireless sensor networks.Aiming at the detection accuracy is decreased because of the existence of faulty sensor measurements, a distributed algorithm for fault-tolerant event boundary detection is advised.The sensors can identify whether they are faulty or not through exchanging the actual readings only one time with their one-hop neighbors.Then normal event sensors detect whether they are on event boundary by statistical comparison.The event boundary width can be adjusted according to the request of network user.This program is of low communication traffic,low computational complexity and low time delay,and thus scales well to large sensor networks.Simulation results indicate the proposed algorithm is of excellent performance even in the existence of large sensor fault probability.