Probability-based fault detection in wireless sensor networks

Abstract

Fault detection of wireless sensor networks has been studied intensively in recent years under the assumption that manager nodes are default probe stations. However, some additional fault detection tasks will make so busy the manager node to be failed more quickly. Moreover, at the first beginning period most sensor networks are working normally without any dead nodes and the probing work are actually ineffective. Thus it is important to study the problems of electing the probe stations and probing frequency. This paper presents a probability-based fault detection algorithm to elect sensor nodes as probe stations by considering the probability distribution of sensor nodes and the fault distribution information (accord with Pareto principle) of sensor networks. The dynamic adjusting rule for probing frequency is also proposed in this paper. The simulation demonstrated that the algorithm and the rule can prolong the lifetime of sensor network only sacrificing very few fault detected rate. The Pareto principle that a small number of clusters contain most of the faults has also been demonstrated, and this principle has been applied to electing probing stations. The results obtained in this paper provide useful guideline to fault management in wireless sensor networks.