Resource Access Control Technology Based On Trust Mechanism

With the advance of computer performance and the development of network technology, Millions of computing resources,data resources have converged in Internet. People require more general resource sharing,cooperative work and joint computing. With its distributed characteristic, Peer-to-Peer has the potential of making resources being more fully utilized, and Peer-to-Peer technology can be used to improve the resource-sharing application.As Peer-to-Peer's altitudinal autonomy, many security problems related to management and access control have arisen. In order to minimize such threats and improve the total efficiency of peer-to-peer resource -sharing systems, the use of trust mechanism is becoming a popular method.This thesis is a part of the research of Peer-to-Peer resource-sharing system. It's object is to implement the secure access control. Based on the study of the existing"trust mechanism", we have proposed a"trust mechanism"including the evaluation of Peer's behavior, the method of"trust information"shared and the strategies of trust. In this trust model, we use probability to define trust degree. In this way, the puzzle of calculating trust degree can be regarded as a problem of time series analysis and forecasting. In a word, our trust model is based on"Statistics"and"Stochastic Process"theory.In order to get the correct evaluation of peers, we use some mature mathematical theory to analyze the Peer's behavior such as Bavesian estimation, stationary stochastic processes theory (AR(p)modal), and queuing theory. In order to get an efficient"trust information"shared system, we add a new method and a public key publication protocol to the old trust mechanism based on recommendation. Furthermore, we have designed a protocol that can not only make the trust information shared more reliable, but also support anonymous search. In order to make our system more intelligent and safe, we have proposed some trust strategies such as a strategy to select reliable file sources based on MD5, dynamic weight decision algorithm based on"Gray System Theory", and the strategy of washing out neighbors.Analyses and simulations show that the model can discard the malicious peer from Peer-to-Peer system effectively and improve the rate of successful dealing greatly. At last the thesis discusses the conclusion and future work.