Research Of Large-scale Media Streaming Video On Demand System Based On P2P

As Internet is widespread in China and multimedia technology based on IP network is developing rapidly, a new interactive multimedia platform, Media Streaming Video on Demand (VOD) System, comes out. Media Streaming VOD is a system that is able to provide interaction between the audience and the programs. The system this thesis does research on provides services to large-scale users, and ensures certain Quality of Service (QoS) at the same time.The thesis firstly researches into the foundation of the system, Media Streaming technology. Its fundamentals, key issues and the shortcoming of current Media Streaming systems are discussed. Traditional VOD systems are based on C/S architecture. The limitation of the I/O capacity of the servers and network bandwidth makes it quite difficult to meet the needs of large-scale users. This thesis presents a solution for this problem, that it, to adopt P2P network in the system, which integrates network resources at a low cost of interaction between peers. After giving detailed explanation of the fundamentals and topology of P2P network, a new way to build a multicast tree on application level is presented. While TCP halves its sending rate when congestion occurs, which arouses large fluctuation in the sending rate, UDP does not have any congestion control mechanism. Therefore, both TCP and UDP are not suitable for real-time media transmission. In order to meet the requirements of the smoothness of real-time media streaming, TCP-Friendly Rate Control (TFRC) is introduced into the system. TFRC flows have a throughput which is similar to TCP flows and enjoys less fluctuation in sending rate, which meets the requirement of the system perfectly. After analyzing the theories of P2P network and TFRC, a detailed design of the system is presented.This thesis researches into a Media Streaming VOD system based on P2P network, and creatively introduces TFRC into the system for the purpose of network congestion control. The system is easy to expand in scale and ensures QoS for the users.