Abstract
Fault detection technique is an important part to be considered for health management of
wireless sensor networks (WSNs). And how to propose an energy-saving fault diagnosis
algorithm is a hot issue in WSNs. In this paper, we analyze the energy-saving strategy in fault
diagnosis, and two energy-saving fault diagnosis algorithms are presented.
Firstly, a two-step fault diagnosis algorithm based on clusters is presented. Cluster heads
are selected through cluster protocols at first. Then cluster heads are initially identified based
on comparisons of data from neighbor nodes because of its spatial similarity, and some
further tests are taken by comparison of each cluster head to get the final state. This could
ensure the high diagnostic accuracy of cluster heads, and the energy in WSNs is saved too.
The simulation results confirm that the two-step algorithm can save almost 30% energy and
the diagnosis accuracy of this algorithm is kept high even though the failure rate is high.
Secondly, a soft fault diagnosis algorithm is given. Sensor nodes are identified by
comparison of data from neighbor nodes because it is spatially related. Considering the
transient fault, a Total Fault-free (TF) model is taken as a reference model, and the algorithm
avoids unnecessary data comparison caused by the transient fault and decreases the number
of redundant detection. Finally, the simulation results indicate that the detection is kept high
and the energy consumption is reduced.
Key words: Wireless Sensor Networks Cluster Network Network Soft Fault
万方数据
