Fuzzy Theory Based Network Nodes Trust Model Study

The size of the network is growing exponentially; a variety of applications in thenetwork becomes more and more popular and plays an indispensable role in people’s life. At thesame time, due to the autonomy, dynamic and anonymity of the trusted network, the stability andavailability of the network was attacked by the malicious nodes of networks and caused greatdamages, the development of the network applications promotion restricted by these unsafe factors.Trust model provides a new way to solve the above problems. By summarizing the literatureof domestic and foreign trust mechanism, found that the current trust models are mostly focus onthe accuracy of the trust value, ignore the fuzziness of the trust itself, this paper, based on theresearch literature at home and abroad, put forward a trust model based on fuzzy theory, the mainresearch content include:(1) Fuzzy comprehensive evaluation based recommendation trust model. To improve thesecurity of transaction in P2P system, this paper proposes a recommendation trust model based onfuzzy comprehensive evaluation. In order to fully embody the fuzziness and dynamic of trustrelationship between the nodes, a two-stage fuzzy comprehensive evaluation method is used tocalculate the direct trust, using the transaction amount and transaction time to determine the weightof second-stage comprehensive evaluation can reduce the artificial error, more accurate. In thecalculation of recommendation trust, using transaction frequency and recommend collusion testingmeasure the recommendation credibility of the recommenders. The simulation experiments showthat the model can effectively prevent the recommend collusion, and has higher success rate inmalicious node detecting.(2) User behavior based fuzzy trust research. In this paper, under the background of thecomplex network we research the use of theory of fuzzy comprehensive evaluation, analysis of thereputation of the assessment strategy based on user behavior, thinking the impact of time aboutcalculating direct trust value, given different weights with the trading in different time periods,making the accumulation of trust with real-time and dynamic; adding "shock penalty factor" in thecount of recommendation trust value to resist shock attack, improved the robustness of the trustmechanism; given a reasonable recommended method and talk about how to choose a moreeffective recommend path. Finally, simulation and performance analysis display that, the trust model can evaluate the trust value more accurately according to contrast with the Beth model, andthe feasibility and effectiveness of this model was proved.