Trust Model Research Based Risk Evaluation

Along with the development of the Internet, the network structure has gradually transformed into the distributed and open mode from the centralized and static one. The traditional security mechanism can not satisfy the new demands of the current network. This problem brings the new challenges to trusted computing. Then, the trust evaluation and management of the network has become the one of hot issues. This paper will apply trust assessment to entity interactions according to the interactions among entities of network. On the basis of the analysis of the insufficiency for the existing trust models, this paper gives a deep research of dynamicity of trust, the relationship between trust and risk and the risk evaluation. The research lays the foundation for credible spatial decision. This paper will mainly study the following works,1. This paper puts forward a new risk evaluation model. In order to make more accurate results of the trust evaluation, this paper considers the elements affecting trust evaluation comprehensively. Through the research of relationship between trust and risk, this paper regards the risk as an important element affecting trust and proposes a risk evaluation model. Before this research, many authors have made the risk as a supplement of trust and calculated it simply, not considering the characteristic of system and entity behavior integrated. This paper analysis the relationship of entity threat events, vulnerability and assets safety value, and proposes a risk analysis framework and calculates risk value according to the framework. The risk is divided into the confidentiality risk, integrity risk and usability risk according to the security features. According to the risk framework, this paper puts forwards Marko forecast method to predict the risk probability. Using this method is not only simple but also accurate to calculate the risk. Comprehensively synthesizing these three kinds of risk value using quantitative analytic hierarchy process can solve the comprehensive difficult problem caused by different quantization of these three kinds of risk.2. This paper puts forward the calculation methods of direct trust and recommendation trust. Trust evaluation involves many other key problems in addition to the risk factor, such as direct trust and recommendation trust calculation. In the calculation of the direct trust value, this paper improves the situation having two interaction results, quantizes trust value in the interval [0,1], and introduces a time attenuation factor to synthesize several trust evaluations during period of time. In recommendation trust computation, this paper puts forward a three-level filtering method to improve the recommendation path and the interactive satisfactory. Finding the credible entity is the premise getting satisfactory services, this paper proposes the concept of recommendation quality trust and service quality trust to find the credibility recommenders. In many other studies, the services-oriented trust and recommended trust always be confused, this usually made the lower recommended trust value entity to be recommended. In addition, this paper also considers the entity similarity and frequent degrees, making the recommendation trust evaluation more accurate.3. This paper puts forward a weighted function of the risk and trust. According to the relationship between risk and trust, this paper designs a weighted function of the risk and trust considering the punishment and incentive mechanism. This paper not only considers the relationship between trust and risk value, and also the effects of the dynamic changes of the risk on trust evaluation. Adding this change factor to the trust computing makes the trust value high sensitivity to risk and can effectively detect and fight against the malicious behaviors. The simulation results verified the effectiveness and high efficiency of the methods presented in this paper.