Research On Cross Domain Trust Model For Internet Of Things

There exist many autonomous domains on the Internet of things,each field has a large number of sensor nodes,most of the existing trust models only evaluate the domain of the sensor node trust,ignore the cross-domain trust,the lack of cross-domain trust envoys point trust as "local trust",trust is only valid within the same domain,node trust cannot be achieved dynamic migration,prevents the node cross-domain interaction.Therefore,the research of cross-domain trust model in multi-autonomous region environment has become one of the key issues that need to be solved urgently on the Internet of Things.Most of the existing cross-domain trust models are centralized architectures,and the trust evaluation algorithm is run by the central server to evaluate the cross-domain trust of nodes.However,these models do not really solve the problem of cross-domain trust evaluation,and there are still some problems such as single point failure of central server,outliers in trust transformation,and trust migration.In this thesis,two cross-domain trust models are proposed to evaluate nodes' cross-domain trust in the multi-autonomous region environment of the Internet of Things.The research contents are as follows:(1)This thesis proposes a three-layer trust model based on fog Computing(CDFC)to solve the problems of cross-domain trust evaluation.CDFC divides sensor node trust into intra-domain trust and inter-domain trust.Fog node in CDFC is responsible for evaluating intra-domain trust of sensor node,and trust server is responsible for evaluating inter-domain trust of sensor node.The fog node uses beta distribution to simulate the historical interaction of sensor nodes,predicts the trust evolution trend of sensor nodes,and obtains its direct trust in the domain.Then,information entropy is used to quantify the weight of recommendation trust given by the recommendation node,and obtains its recommendation trust in the domain.On this basis,the direct trust and recommended trust of the sensor node are fused to obtain the in-domain trust of the node.Finally,the trust server obtains the inter-domain trust of nodes through weighted domain trust and intra-domain trust of nodes,and realizes the evaluation of inter-domain trust of nodes.CDFC reduces the computing pressure and network delay of trust server and reduces the security risk of trust evaluation by introducing fog computing.Compared with the Domain partition-based trust model(DPTM),CDFC uses the information entropy quantization method to calculate the weight,so the trust evaluation accuracy is higher and the evaluation result is closer to the real trust of nodes.(2)Although the multi-level architecture solves the problem of cross-domain trust assessment,there is still a defect that the inter-domain trust of sensor nodes is affected by the behavior of all nodes in the whole domain.Therefore,a cross-domain trust model based on normal distribution(CTN)is proposed.CTN firstly constructs fuzzy inference rules and membership function between the success rate of interaction and the trust relationship of nodes through fuzzy set theory,and obtains the historical interaction knowledge of nodes.Then filter malicious recommendations and compute nodes recommend trust.Finally,the cross-domain trust of nodes is obtained by combining historical interaction knowledge and recommendation trust.In the calculation process of recommendation trust,CTN also creatively uses the central limit theorem to calculate the normal distribution parameters corresponding to the mean value of recommendation trust,and removes the outliers in recommendation trust through the transformation of normal distribution parameters.Compared with the cross-domain Trust Evaluating of IoT(CTE),CTN considers the outliers in Trust transformation,can normally convert recommendation Trust,and reduces the Trust loss caused by node migration.(3)The cross-domain trust management system is developed and implemented.Based on the requirement analysis,overall design and module design of the cross-domain trust management system,two proposed cross-domain trust models and three existing intra-domain trust models are developed and implemented on the basis of Java Web development framework Spring Boot.