Study On Unstructure P2P Network Of Trust And Reputation Mechanism

With rapid prevalence of the application on unstructured and distributed peer-to-peer technology which is based on Gnutella protocol people have been enjoying it on the convenience of using resource-sharing. However, malicious nodes in P2P network can easily make use of this technology characteristic to modify or discard messages routed at will; personate hot information with vicious codes such as Trojan horse for other nodes download; forge a great many of bogus identities or unreachable messages to launch DDos attack aiming at some nodes or the whole network.The cause of these problems stems from the initial design that the Gnutella P2P network assumes that peers have confidence in one another and resource-sharing among them is trustful and reliable. In fact, the trustful relationship among peers is unexistence so that the malicious behavior can be done by using this assumed trust relationship among peers. Consequently, it is important for peers in the network to construct a mechnisam of trust-and-reputation. The construction of trust-and-reputation mechanism among peers can provide a way of evaluation for choosing trustful peers and reliable resource-sharing. However, existing researches at this field have not made any acknowledged conclusion in the mass, and the focus on the mechanism is mainly studies and improvements of these trust-and-reputation models. That is to say, it is necessary for constructing secure P2P systems and extending P2P applications to research the trust-and-reputation mechanism among peers.Analysis of peers' value of trust-and-reputation, the contents of trust-and-reputation mechanism for Gnutella P2P are proposed, which mainly deal with the value of the trust and reputation computing model, the secure transport and the policy of secure storage. Moreover, based on systemic studies for these above contents, a few novel achievements are obtained and listed as follows:(1) A trust-and-reputation model based on the Markov process of describing the peers'. dynamic behavior is proposed. Moreover, according to the random characteristic of the peers' behavior process and ones' state results, a new-style trust and reputation model forecasting and reflecting the peers' dynamic behavior process is constructed.(2) A partition and multi-path transmission protocol of the peers' trust-and-reputation value is proposed. Based on the analysis of the peers' trust and reputation value concerning their integrality during the process of transmission, the Shamir secret threshold scheme is introduced into the protocol. In the protocol transmission, the peers' trust and reputation value are divided into segments and then resented back through multi-path. The method is provided with a new substitute for. security transmission of the peers' trust-and-reputation value.(3) A redundant secure storage policy of the peers' trust-and-reputation value is put forward. In the policy, according to the requirement of the peers' trust-and-reputation security storage, Cauchy code theory is introduced into the content of the peers' trust-and-reputation distribution and storage to build an algorithm of redundant code. Since segments distribution is held among trusted peers, and those fragments are scattered ones, storage nodes can make nearly no knowledge of the local conserving information. This way can excellently avoid malicious peers' modification on them so that the validity and security on storage of trust-and-reputation value are enhanced.