The Trust Management Research In Peer-To-Peer Networks

As the most up and coming technology, the peer-to-peer network technology (P2P) is increasingly receiving more and more attention from the academic and industrial circles, and is considered one of the most important technology that lead to an evolution of the web in the future. The P2P network broke the traditional model based on C/S structure, and brings lots of convenience for people’s information contract, as well as some trust absence due to its openness, autonomy and dynamics natures. However, the traditional security mechanisms, including information encryption, identify authentication, PKI technology which based on role controlled can’t meet the need of security recently, and has greatly hampered the development of P2P systems. As one of the most effective ways to solve the trust absence appears in the P2P network, how to establish trust management systems is the key to ensure the security exchange.With the development of research, some comprehensive trust models have been built and give lots of good theoretical support for the trust problem solving. But the initial trust, the dynamic description and the dealing of the recommendation path still under the way to be solved. Aiming at these issues, a trust management model for dealing with a complex recommendation relationship was proposed to realize interactions between strange nodes in this paper, and the model mainly including the following four aspects.Firstly, a method computing the initial trust value was presented and meanwhile its application scope was given, namely only useful for a newly node. Once there is some transactions information appears, the initial trust will lose its utility. Secondly, two important definitions named query similarity and honesty were raised and then combined with the historical information to compute the direct trust value. This model can improve the query’s performances of resource and the consistence of nodes’languages and behaviors. Thirdly, the statistical characteristic of the nodes’trust value was used to compute the recommendation trust and then the fuzzy analytic hierarchy process (FAHP) method was introduced to compute the weight factor in the synthesis of trust value. Finally, a new method for dealing with the complicated trust graph was proposed to simplify the complicated trust graph and avoid one link being used more than once in trust computation. Experimental results show the efficiency of the model.In addition, aiming at the problem that the relations of nodes’trust value in the open network change dynamically with the time and transactions, a update model is proposed. And through studying the principle of memory storage, update, forget and the memory characteristics of the social network, we consider that the importance of nodes’service, the influence of events and other attributes are kinds of information with memory, and the trust information of nodes with these attributes in the network continuously changes with time going by and transactions taking place. On the foundation of the cumulative characteristics of memory, this paper raised a trust value update model which decay over time, and have a fading factor, incentive mechanism. So that it can describe the evolution process of the trust value more clearly. The effectiveness of the proposed model is validated by the experimental results. At the same time, it can achieve detecting the malicious nodes by monitoring the frequency of transactions.