Tree Topology Based Fault Diagnosis in Wireless Sensor Networks

In order to improve the energy efficiency of the fault diagnosis in wireless sensor networks, we propose a tree topology based distributed fault diagnosis algorithm. The algorithm maintains high node fault detection rate and low fault alarm rate in wireless sensor networks under low node distribution density. First, the algorithm finds a good node with a multi-layer detection method, then, detects the status of other nodes by means of the status relation with the good node which is inferred by the parent-child relation in the tree topology, and up to achieves fault detection of the whole network. In ZigBee tree network, the simulation results show that the algorithm performs better in energy efficiency, and identify the faulty sensors with high accuracy and robustness.

I. Introduction

Wireless sensor networks (WSNs) consist of lots of small nodes which have communication capacity and computational capability. WSNs are deployed in a specified area to collect interesting information and send it to observer for processing and analysing (in [1]). Because the low cost sensor nodes, which computational capability, storage capacity, wireless communication capacity and the energy is limited, are deployed in uncontrollable and hostile environments, so it is common for faults to occur due to component failure, battery depletion and some unknown reasons. The faults will reduce the reliability of the network and dirty data collected by the faulty nodes will influence the analysis result of the whole network (in [2]). So, the fault detection is an important problem in WSNs research, and we will pay close attention to fault detection method in WSNs.