Network-friendly P2p Application And Performance Optimization

Applications relying on peer-to-peer (P2P) architectures have become massively popular:file-sharing, video streaming (IPTV, VoD and time-shifted IPTV), grid computing, WEB service, and network game. The fundamental idea of P2P network is to have peers cooperate in an overlay network and operate as both servers and clients, and then the service burden is distributed to all participating peers from the bottleneck servers. Indeed, comparing with the traditional Client/Sever (C/S) model, cooperation between peers in P2P model is commonly implemented with multiple concurrent TCP connections in file-sharing applications as well as in most recent video streaming systems. The aggressiveness and robustness of P2P technology remarkably improve transfer efficiency and network bandwidth utilization. However, P2P applications consume most of Internet bandwidth and bring about congestion in access network of Internet Service Providers (ISPs), which severely deteriorates quality of services. P2P technology florishes new applications in the information era of broadband Internet network, whereas ISPs lose control of these P2P applications. Four perspectives are of special interests of this dissertation.(1) We develop a performance modeling system of P2P file sharing traffic and traditional Internet traffic (WEB traffic) which quantitatively analyzes the impact of P2P file sharing traffic on WEB traffic in the congested access network. We answer the following questions from a user's point of view:How many P2P concurrent connections and what proportion between P2P traffic and WEB traffic will guarantee WEB performance. Simulation on NS2 simulator finally validates our model.(2) For file-sharing systems, we propose a friendlyP2P system with new application-level approaches for congestion detection and avoidance to keep fairness between P2P traffic and traditional Internet traffic. The friendlyP2P, which is friendly to ISPs, namely to Internet networks and traditional Internet traffic, detects network congestion via throughput measurements and alleviates network congestion by optimization of the number of P2P connections from the viewpoint of P2P users. Compared to a basic P2P application, friendly P2P can improve the performance of traditional Internet traffic when network congestion occurs.(3) We further tackle the issue of efficient channel switching in P2P IPTV systems. First, we conducted and analyzed a set of measurements of one of the most popular P2P systems (PPlive). These measurements reveal that the set of contacts that a joining peer receives from the central server are of the utmost importance in the start-up process. We then formulate the switching problem, and propose a simple distributed algorithm, as an illustration of the concept, which aims at leveraging the presence of peers in the network to fasten the switch process. Finally, simulations show that our approach leads to substantial improvements on the channel switching time.(4) One of the most attractive feature of IPTV is the so-called catch-up TV. We emphasize the challenging characteristics of catch-up TV systems that prevent known current IPTV and VoD delivery systems to be used. Then, we formulate the catch-up TV as an interval graph. Based upon this modelling, we further present a peer-assisted system, namely PACUS, where a set of end users'computers assists the server for the content delivery. We show in particular how the PACUS tracker server can be efficiently implemented for catchup TV. We demonstrate the benefits of PACUS by simulations. We especially highlight that PACUS reduces the traffic at the server side with the advantages of lightweight and self-adaptive unstructured peer-to-peer systems.