P2P-Based Trust Mechanism Designs

There are a lot of difficult problems in P2P Field, such as efficiency, safety , trust and reliability etc. This paper focuses on the trust model research and aims at designing and implementing a trust model which has high efficiency, reliable safety and low cost, etc. It will reward those peers who provide good services to others, and punish those who provide bad ones. And it will also force those classic malicious behaviors gain no profit.There's a method called Maximum Similarity Estimate(MSE) in Probability & Statistics. We can use MSE to estimate the probability that peers provide good services based on their past behaviors. As MSE comes down to a mathematic problem that finding minimizer of single-variable function. This problem should rely on a powerful tool– MATLAB. This paper use VC++ .NET to obtain MATLAB's ability to solve that mathmatic problem and analysis the results. At the meanwhile, in our real lives, people who have same or similar interests tend to form a group or so called fellowship. Fellows tend to trust and cooperate with each other with high degree, and they have quite a lot privileges.This paper makes a trust model called FTrust (presents for Fellow/Friend Trust) out of Fellowship and MSE, and gains some brilliant outcomes. The routing part of FTrust adopts Kademlia, a splendid DHT arithmetic that has been used in many file sharing P2P applications such as eMule, BitTorrent, and BitComet, etc. It advances FTrust's efficiency greatly.Simulations show that FTrust can converge within a considerable time, and gain great results. FTrust has some unique superiority when compared to MSE's simulation, what's more, it can restrain collusions which most of the Trust Models existed cannot.