Research On Fault Management In Wireless Sensor Networks For Node Energy Consumption

Wireless Sensor Networks(WSNs) nodes have the features of constrained energysupply, limited hardware resource, large numbers, self-organization and dynamictopology, which make networks prone to have malfunctions. It will not be able to avoidthe faults completely though there is redundant fault-tolerance. The energy fault causedby node energy depletion is widespread typically. Thus, in terms of network faultscaused by failure nodes, this thesis studies fault prevention, fault detection and faultrecovery in WSNs, and proposes corresponding methods based on reducing energyconsumption and balancing energy consumption. Here, the research content mainlyincludes the following five aspects:(1) For the reason that the redundant routing method for fault prevention in WSNshas the shortcomings of too much traffic and more wasted energy, a Fault PreventionTechnique based on Controlling Redundant Routes into Sleeping (FPCR) is proposed.Sink node which collects each node’s residual energy and number of neighbors, andcalculates energy consumption degree parameter and health degree parameter, which areused to choose the opptimal routing by A-Star algorithm. And then, it controls theredundant routes into sleeping mode in turn. In this way, the purpose of preventingcertain node depletion in advance and prolonging the networks’ lifetime should bereached. The simulation results of NS2and actual node experimental results illustratethat the presented health degree based technique can balance the network’s energy andprolong the network’s lifetime significantly.(2) Because of the physical space distance, the sensing data have distinguisheddifferences between neighbors in neighbor coordination method for fault detection inWSNs, which will influence the fault detection accuracy. To solve this problem, A nodeSelf Detection Technique by History data and Neighbors (NDHN), and meanwhile, aFault Detection Technique based on Clustering (FDTC) is proposed to reduce the falsealarm probability of NDHN. It weights the nodes’ historical data and the neighbors’sensing data in NDHN, and run the judgment task in each node. Then, by usingmathematical clustering algorithm, the sink node clusters each node by its sensing data.Employing carefully-chosen threshold, fault detection task is implemented on the sinknode. The simulation results show that the presented method can obtain high faultdetection accuracy and low false alarm probability, while greatly reduce nodes’ energyconsumption.(3)“Coverage hole” caused by failure nodes is a typical fault in WSNs, which isvital to evaluate the quality of networks coverage. To recover the “coverage hole”, aRecovery Algorithm based on Minimum Distance Redundant Nodes (MDRN) is proposed. The recovery algorithm is implemented on the sink node with unconstrainedenergy consumption. The locations of all active nodes and redundant nodes, known bysink node, are used to calculate the “coverage hole”. Then, sink node calculates theoptimal location to recover the “coverage hole”, and searches for the minimum distanceredundant node from which as the alternative node. On this basis, a Method to Estimatethe Network Coverage based on Monte Carlo Integration (ECMC) is proposed toevaluate the quality of coverage of MDRN, by using probability sampling method toestimate the area of hole’s irregular region. According to the experimental data, theexperience formula of network lifetime and the ratio of redundant nodes to active nodesis constructed, and in the same way, the experience formula of network coverage andthe ratio of redundant nodes to active nodes. Those provide theoretical reference inchoosing reasonable number of deployment nodes at the beginning of the networkdesign, in accordance with the monitoring area, expected monitoring lifetime andnetwork coverage. Simulation results demonstrate that, by choosing appropriate numberof redundant nodes, this algorithm will be feasible and efficient for dealing with the“coverage hole”, simultaneously get great recoveryaccuracy and network coverage, andalso achieve the purpose of prolonging the lifetime of WSNs. In the case of guaranteesufficient sampling, the calculation error of network coverage can be controlled within5%, where the program is simple and the calculation result is believable.