The P2p Incentive Mechanism Based On Game Theory

Recently, with the advancement of Internet technology, people have been paying more and more attention to Peer-to-peer (P2P) networks and P2P document sharing is the most important application of P2P technology. As the most pervasive application in P2P document sharing systems, BitTorrent document sharing system can distribute mass files in a quick and effective way while not burdening source servers. However, in practical applications, many participants don't have incentive mechanisms and they are unwilling to share resources with other users. This leads to free-riding problem, which is the common issue in traditional BT systems, and greatly impedes the development and application of document sharing technology.Through the experiments, this thesis shows that there is free-riding problem in current BitTorrent document sharing systems and points out the cause lies in the Optimistic Unchoking Algorithm in BitTorrent systems, which makes free-riding nodes obtain a lot of service resources from seed nodes and access nodes. In order to solve the free-riding problem, this thesis proposes MCS (Maximum-Contribution-Satisfaction) bandwidth allocation strategy, which defines a profit function based on nodes'contribution and satisfaction. Besides, this thesis combines constraint conditions with object function to construct the constrained optimization problem of seed bandwidth allocation and proves that there is system optimal solution to the problem. According to the constrained optimization problems, this thesis constructs Lagrangian function and finds a bandwidth allocation strategy by using Kuhn-Tucker conditions. In the end, this thesis proves the existence of Nash equilibrium by analysis of bandwidth allocation strategy and works out a solution to Nash equilibrium.The experimental results demonstrate that there are less free-riding nodes but more contributing nodes in BitTorrent document sharing systems using MCS algorithm and as time goes by, the average downloading time of free-riding nodes increases while the average downloading time of contributing nodes remains or even decreases. The results also show that the bandwidth utilization of seed node is significantly improved in the new systems, which further proves that the bandwidth allocation strategy of MCS algorithm can effectively eliminate free-riding and improve the system performance.