Research On Reliability Evaluation Method For Wireless Sensor Network

The reliability of wireless sensor network(WSN)is the key for ensuring the normal operation to improve the performance of network.The reliability is the main point of WSN network design,and it is regarded as an important basis for ensuring the normal operation and maintenance of the network.Due to the specificity of WSN and work environment,the network structure is significant different from WSN and traditional wireless network.In WSN,because of the limitation of the sensor's individual computing power,storage resources,battery capacity,bandwidth limitation,low transmission rate,mutual interference among signals,the transmission signals are constantly reduced with the extension of communication distance.Moreover,the sensor is easily impacted by the factors of bad weather and electromagnetic radiation.Also,WSN network is vulnerable as the network attacks,such as passive eavesdropping,active intrusion,and denial of service.Based on the above factors,with the study of WSN reliability,it is necessary to comprehensive consider different factors impacted on the WSN in order to achieve the fully evaluation of WSN reliability.The reliability evaluation of WSN is a difficulty in the field of academic research,and to date it is still unable to form a complete and mature theoretical framework.Therefore,this study has an important theoretical and practical value.The research of this paper is mainly focused on the following aspects.In view of the design problem of reliability evaluation framework of WSN,the main influencing factors are summarized through the analysis of various links that affect the reliability of WSN.The factors are divided into internal and external factors.Based on the effect of different factors on WSN,the WSN reliability evaluation index system is constructed,where the WSN reliability evaluation is completed by evaluating the WSN fault status and evaluating the WSN security status.On researching the WSN fault problem,it mainly studied the problem of WSNnode faults.On researching the security of WSN,it mainly focuses on the WSN network attacks.At the same time,for the multi-factor reliability evaluation of WSN,a hierarchical reliability evaluation model is designed.Based on the above research,a framework for the research of WSN reliability problem is established,which is easy to be expanded and can realize the reliability evaluation of WSN.Aiming at the problem of node fault diagnosis in WSN,a WSN fault diagnosis method based on belief rule base(BRB)is proposed by analyzing the data of sensor nodes collected from the data processing center.BRB can effectively use quantitative data and qualitative information.It can effectively describe the knowledge of fuzzy uncertainty and probabilistic uncertainty.It has a powerful complex system modeling ability.As such,in the research of fault diagnosis,first,the time correlation,space correlation and attribute correlation among sensors are analyzed.Second,the clustering method is used to cluster the sensor nodes in WSN.By comparing the data between the nodes in the cluster,the abnormal data are found and the fault detection of the WSN is realized.Third,the fault diagnosis model is constructed by hierarchical BRB,and the reasoning process and parameter optimization method are distributed.Finally,the experimental results show that this method can effectively identify all kinds of WSN faults,and has a better accuracy of fault type detection compared with other methods.Aiming at the problem of network intrusion detection in WSN,by analyzing the characteristics of network connection data in WSN,a WSN intrusion detection method based on convolution neural network(CNN)is proposed.CNN is the most advanced technology in the field of deep learning.It has a strong ability in mass-data processing and feature learning.It is especially suitable for processing data with high-dimensional attributes such as text,audio,image and video.Therefore,in the research of intrusion detection,first,a CNN-based network intrusion detection model is constructed according to the network characteristics of the WSN.Second,to make full use of CNN's characteristic learning ability,a graphical processing mechanism for data is designed to convert the network connection data into grayscale and RGB maps to adapt to the useage of the CNN model.Third,according to the requirements of intrusion detection,the appropriate CNN model is selected to introduce the idea of transfer learning to model training,and the implementation process of the network intrusion detection model is obtained.Finally,it is proved by experiments that thismethod can be effectively applied to WSN network intrusion detection,and has high detection accuracy compared with traditional detection methods.Aiming at the problem of WSN reliability evaluation,by analyzing the internal and external characteristics of WSN reliability,a WSN reliability evaluation method based on hierarchical BRB is proposed.Firstly,a hierarchical BRB model based on WSN fault evaluation,WSN security evaluation and WSN reliability evaluation is constructed by using expert knowledge.Secondly,based on the iterative ER algorithm,the model reasoning process of WSN fault evaluation and WSN security evaluation without sample training is given.Based on the analytic ER algorithm,the reasoning process of WSN reliability evaluation with sample training is given.Thirdly,the parameters for the model are optimized with samples to obtain more effective evaluation results.Furthermore,it is proved by experiments that the method can effectively evaluate the reliability of WSN,and the accuracy of the evaluation is better than the traditional methods.Finally,through the analysis of real engineering cases,the feasibility of the method in engineering application is proved,which indicates that the proposed method has a good application prospect.