Transmission Control And Incentive Mechanisms For P2P Networks

Peer-to-Peer(P2P) has become the overwhelming technology for data sharing, distributed storage, cooperative working and large scale parallel computing. Assuring the cooperation between P2P and other network elements in the Internet, and try to stimulate the cooperation of peers in the P2P network are important issues deciding the system performance and the future development of P2P. This dissertation studies the problems of transmission control and incentive mechanism of the P2P networks, and results in the following achievements:(1) We propose a low priority transmission control algorithm TCP-CPLP(Congestion Pricing Low Priority TCP) for the P2P flow in order to mitigate the conflict between P2P applications and traditional network applications. TCP-CPLP can adjust a connection's sensitivity to network congestion according to the congestion price the user sets for that connection, thus achieve multiple flexible priority levels of service between standard TCP and absolute low priority TCP. Simulation results prove the fairness, stability and validity of the proposed algorithm.(2) We propose a novel incentive mechanism that aims at reducing cross-ISP traffic as well as stimulating sharing. We bring the role of ISP in the currency-based incentive, where the incentive mechanism uses virtual currency to prevent free riding and stimulate sharing, at the same time charges a transmission cost for inter-domain traffic. Under such incentive mechanism, the download peer pays the upload peer for the sharing service, and additionally also pays the transmission cost if it incurs any cross-ISP traffic. Simulation results indicate that the ISP-friendly incentive mechanism can significantly reduce the cross-ISP traffic as well as improve sharing and decrease free-riding.(3) Users in a Peer-to-Peer Video on Demand (P2P VoD) system are allowed to watch any video file at any point in time and to use VCR operations which decreases their ability to help each other and offload the server. So we propose a lightweight incentive mechanism based on virtual currency to improve the performance of the system. If a peer shares a video file which he has completely downloaded before, he will get paid by the audience peers who are in the process of watching that video. Game theoretic analysis proves that the proposed incentive mechanism can significantly relieve server's heavy load and increase users'contributions. Simulation results prove that the incentive mechanism can improve sharing between peers and reduce the load of server.(4) We use game theory to analyze the incentive mechanism of private P2P network. The private P2P network uses a upload/download ratio-based incentive which significantly increases the sharing of peers in the P2P system and get a higher performance compared to the public P2P networks. We analyze the production, sharing and downloading of the peers, and model the system with a sequential game, get the condition of the system parameters when the network achieves a sequential equilibrium, which can be used as a reference to set proper system parameters.