Research On Node Soft Fault Detection And Processing Technology Of Wireless Sensor Networks

Wireless sensor networks(WSN)are used to perceive the objective physical world and acquire the information volume of the physical world,which has broad application prospect.However,the cost of the sensor node is cheap,but the deployment environment is complex,resulting in soft fault of the sensor node.The node can still send the measured value,but the measured value deviates from the actual value and reduces the reliability of the WSN.In this thesis,for the soft fault detection and processing technology of sensor node in WSN,the following three aspects are researched:(1)Firstly,for the defect of traditional node fault detection algorithm: large amount of traffic or the need of prior knowledge,a node fault detection algorithm based on spatiotemporal correlation is proposed.The node first performs self-detection and uses time correlation to judge the current measured value is connect or not bared on historical data.If it is possibly abnormal,the similarity detection is performed.The abnormal node can get its state by majority vote or the result of similarity detection with state determined node.The simulation result shows that this algorithm has higher fault detection accuracy and lower false alarm rate.And the node first performs self-detection,which can reduce the communication load and energy consumption,so as to prolong the service life in wireless sensor networks.(2)Secondly,in the light of the problem that the previous data aggregation algorithm has large computation load or requires prior knowledge,a data aggregation algorithm based on autoregressive model is proposed.Firstly,the autoregression model is used to fit the measured values,and calculates the predicted value of next measured value.Use the predicted value and the actual value to calculates the measurement variance according to the orthogonal principle of linear minimum mean square error estimation.Use lagrange multiplier method to calculates the optimal weighting coefficient,and then obtains the aggregation result.Because there is no need of prior knowledge,the algorithm has good adaptive ability.The simulation result shows that this algorithm not only has less error,but also can better process the measured value to get the aggregation result closer to the actual value,In addition,the performance of the algorithm is more stable.(3)Combined with the node fault detection algorithm based on spatiotemporal correlation and the data aggregation algorithm based on autoregressive model,the node soft fault detection and processing prototype system is designed and implemented.The temperature sensor is used to collect the ambient temperature value,the state of the sensor is judged,and the sensor measured values in the monitoring area are aggregated.By comparing with the situation before the improvement,the node soft fault detection and processing system proposed in this thesis can judge the state of node with better accuracy and can get the more accurate ambient temperature value by the aggregation process.