P2P Trust Model Based On The Time Window And Risk Value

P2P is a peer-to-peer network,and now it has ben one of the most important members of theInternet application systems. Due to the properties of openness,anonymity and loose coupling, there issome serious safety issues—for example,having no faith--between the users in the Internet systems.Therefore, it results in a large number of service quality problems, which is not reliable, such as theFree-Riding issues, Tragedy of Common isstes and this kind of selfish behaviour; and lots of fraud andforgery hostile behavior, etc. As a result, how to build trustniess between users in P2P systems, removingmalicious behavior and improve the service quality has become an urgent issue needed to be resolved.Trust and reputation model have been an effective way to overcome the difficult problem.This article is mainly about the current issue of time attenuation of trust and reputation system,and the problem was deeply discussed in the paper.1.In the base of current research about trust and the model of reputation development, it putforward timewindow accurate factor. By dividing the timewindow into the different cycle and time, itquantifiys the time function, and also achieves the purpose of detailed recording the time when it happens.Due to the great impact transaction records on reputation value differs at different time, therefore, thereputation value may become invalid since the time attenuation; and the trust value achieved with the factorwill be more accurate and reliable. In this way, it can make sure the success rate of transaction and thesecurity of network. Finally, by simulation it proves that as the growing volume of transaction cases, thesuccess rate of the trust models with the timewindow factor will be higher than the original model.2. Since most of the malicious transaction behaviours in the current trust model are testing afterthe event, the paper combines the refined the concept of risk in economics—the refining risk value, and putfoaward the risk predictors. By dividing the risk identification and evaluation into different levels, itestablishes the risk function. Risk predictors can be refered as the reference factors of predicting the futurebehavior of node, and it can also be combined with the reputation value to account the trust value of node,meantime, it can be recognised as the punishment of the malicious trading behaviour and the identification ways of the malicious node. Then by simulation, it demonstrates that the trust model with risk predictorscan still keep high transaction success rate at the condition that the malicious node continuous increases.3. The paper put forward the timewindow and risk value based trust model at the foundation oftimewindow accurate factor and the risk predicto（rTimewindows And Risk Trust,TAR-Trust），and finallyby simulation, it demonstrates that the trust model can still has better performance in the case that differenttypes of node attack at different degree.