Research On Traffic Optimization Of P2P Live Streaming Service

In recent years, P2P (Peer-to-Peer) technology emerges as a solution that can solve large scale content distribution problem efficiently. P2P networks can guarantee low cost and high extendibility. Every peer in P2P network acts as both consumer and provider for a specific resource. Peers construct a large user resource pool by sharing their resources to accomplish a global task which only needs the support of relatively small infrastructure.As a dynamic and real-time P2P service, P2P live streaming service draws more and more attention from people. In order to achieve the requirements of live streaming service, network traffic must be optimized. Existing optimization schemes of P2P live streaming service have some shortages. Such as fail to consider the stability, nodes join and leave system frequently may cause instability of the system; traffic localization causes traffic over-concentrated and increases the risk of network congestion; generating a lot of redundant traffic and overhead in the process of data scheduling.Aiming at the shortages of existing schemes, this thesis proposes a P2P live streaming traffic optimization scheme. In the process of choosing neighbors for every peer, the scheme takes the nodes’online time into consideration to make the neighbor selection prefer the peers with longer online time. In order to prevent network congestion, monitoring the load rate of peers and PIDs (Provider-Defined Network Location Identifier). When a peer’s load rate reaches the predefined threshold, it will be no long assigned as a neighbor peer. What’s more, the server checks the load rate of PIDs periodically. In order to reduce the redundant traffic and overhead, this thesis proposes that the buffer maps of peers will be managed by a server in a centralized way, and the server will record the request times of every peer, then forwards the data request to the most appropriate neighbors.First, this thesis analyses the features and shortages of many existing P2P traffic optimization schemes. Then, it describes the neighbor selection mechanism, congestion control and data scheduling scheme proposed in this thesis with several chapters. At last, it shows the experimental results based on the OMNeT++simulation environment and analyses about the results.