Technology Research Of Transmission And Schedule Of Real-time Multimedia Streaming In Peer-to-Peer Network

Media streaming will become the mainly driving force for future broadband network application. But media streaming imposes a high load on the underlying network and the streaming server, and there still exists many difficulties for deploying it in a large-scale. Almost all of the existing works on multimedia streaming is based on client-server models. Server network bandwidth runs out rapidly if unicast client-server model is used. Single source multicast is one of the solutions that use a single stream to feed all the clients. The deployment of IP multicast has been slowed by difficult issues related to scalability, and support for higher layer functionality like congestion control and reliability. CDN (Content Delivery Network) push the content and service to the Internet edge, which could alleviate burden on media servers and network infrastructure, but normal ICPs could not afford the expensive charge.A peer-to-peer(P2P) model is ideal as a model to solve the server link bottleneck problem. In the P2P model, multimedia contents are distributed by using the bandwidth of the clients themselves.The clients in P2P systems contribute resources to the community and in turn use the resources provided by other clients. Data traffic is not localized on a specific site since the peers cooperate for sharing contents. It is typical that there is no central server that holds the contents, and peers work on an equal footing. There has been much work on P2P systems in recent years. Those works dealt mainly with data lookup and storage management and application level multicast in a general P2P system. However, It can be said that the work on P2P media streaming systems is still in the early stage, and there is still some room for improvement of performance and generalization of model.In the paper, we focus on two problems as follow: (l)the transmission scheduling of the media data for a multi-supplier streaming session in P2P media streaming, and (2)peer selection in P2P real-time media streaming. To the problem (1), we study and analyse deeply two existing media data assigning algorithm OTS and FSS , and further propose three media data assigning algorithms, in which, ZBS algorithm is better adaptability and lower buffering delay than OTS and FSS, and alleviates the overlead maintaining taking over streaming in receiver. The simulation results demonstrate that ZBS algorithm results in accelerating speed at which the system's streaming capacity increases. To the problem (2), we study deeply a novel P2P real-time media streaming service called CollectCast, and point out the faults of peer selection problem model and its algorithm in CollectCAst. We modify or improve them. The simulation results demonstrate that the improving scheme results in higher quality streaming in receiver.