Research On Fault Detection Algorithm Of Wireless Sensor Networks

Wireless sensor network(WSN)is a kind of wireless data acquisition network,and consist of special deployment in a number of specific work environment,mobile or stationary sensor nodes through self-organization,distributed form collaborative data acquisition,analysis integration,report,etc.WSN is usually used in bad,dangerous environments,data collection and so on cause which with the flexible,easy mobile and extension etc.In most sensor network environments,the reliability of transmission between nodes is not the primary goal,but the reliability of data transmission is the first.A large number of resource-constrained nodes are included in the sensor network,and the information transmission between nodes is completed through multi-hop communication.Precisely because of these characteristics,the failure probability of nodes in the network is higher than that of traditional networks,so it is necessary to establish a fault detection mechanism for sensor nodes.In this paper,combined with the characteristics of modern wireless sensor communication network and research of clustering idea,and based on the existing wireless sensor network fault detection algorithms we proposed a fault detection algorithm of sensor networks based on SVR prediction,construction of sensor node deviation level model,and node deviation level update mechanism.In the face of large-scale wireless sensor network,environment,the network is divided by clustering.In the process of clustering,an improved LEACH clustering algorithm is adopted,and an energy model is introduced to describe the performance of the clustering network,so as to make clustering more uniform,balance the energy consumption of nodes and reduce the overall energy consumption of the network.Member nodes in the local executive SVR prediction algorithm,according to the data with prediction and compute nodes level of deviation.And members in different deviation grades send the forecast information of sample data with different frequency.The cluster head detects the fault and distinguishes the node faults according to the standard set by the stored deviation data set with transmitted data.At the same time,the selection of kernel function is simulated and analyzed in the prediction model.Simulation results show that the SVR predictive fault detection algorithm maintains excellent performance in term of fault detection effect,fault detection rate,detection accuracy and detection energy consumption in wireless sensor networks of different scales.Figure [51] table [8]...