Research Of Multi-level Trust Calculation And Trust Transmission Mechanism In P2P Networks

With rapid spread and popularity of networks, trust has become one of the key issues in internet transactions. Appropriate designing of trust model system acts as an important role in each interaction via internet. However, there are still many shortcomings in the trust model of P2P network: First of all,current trust model approach is too simple which limits reasonable measurement of trust between entities. Secondly, few research works have been done to address trust transmission. So the above problems multi-level node-based Trusted Computing program and the similarity of the trust transfer methods have been proposed in this thesis.In the multi-level and role-based trust evaluation scheme, through the separation of different roles for each entity and the classification of evaluation factors integrated with the introduction of evaluation weight. According to atttibute has different levels impact of interaction,evaluation factor for classification.According to the contents of each interaction effect of different levels of trust, each evaluation is given on the corresponding evaluation weight.Put forward the concept of the role of independent, each node acts as a role in the evaluation value can only be used for the calculation of the node to act as the role of trust. Gives an independent calculation scheme recommended trust.In the calculation of the credibility of recommended uses the similarity of behavior and dynamic updates as the way, so that credibility of the recommended is no longer attached to the credibility calculation.Delivery methods based on similarity of trust, which is in the absence of direct interaction but in the case of direct recommendation, transfer the trust by calculating the similarity of the recommendated nodes. In the case of no direct recommendation, firstly confirm the chain of trust effective, by the Trust Guarantee Chain, then recommendate of the trust and complete the transfer. At the same time, for most of the trust models ignore the issue of multi-step recommendation, the nodes which are responsible for the delivery of trust are devided into Guarantee Nodes and Recommendation Nodes. Recommendor nodes are used to calculate the similarity, and Guarantee nodes are used to guarantee the chain of trust availability.Introducte the calculation of the node similarity, and put forward the concept of Trust Guarantee Chain. Node attributes also be refined to distinguish. The concept of Requirement Similarity and Evaluation Similarity are proposed, and put forward the concept of tuple recommendation trust.Finally, the multi-level trust calculation program used in the P2P file transfer system, the experimental data show that the confidence level calculation method can greatly improve the success rate of transactions more effectively to establish a trust relationship between nodes. The simulation results also prove that the trust based on similarity degree transfer program known especially for small systems, and to a large extent guarantees the reliability behavior of network entities.