Theoretical Analysis And System Implementation On Peer-to-Peer Media Streaming

As the development of computer science and Internet technology, net resources are increased dramatically. How to realize the distribution of immense data quickly and efficiently has become a key problem that people are facing. Thus, Streaming media distribution platform based on the Peer-to-Peer Network came out. It has already become the core problem of distributed system research.This thesis firstly researches into the Peer-to-Peer Network and the Streaming media technology. It analyses core technologies about Application-Layer Multicast and the infrastructure of Peer-to-Peer Network, its topology, the development of Streaming media and associated key technologies. According to the limitation and disadvantage of the existing accomplishments, a new overlay model TBMM based on the mixed network topology is presented. It aims at realizing a mixed network topology with high throughput, balanced load, less network diameter, high fault- tolerance and high amplifying capability.Then, the nodes in the Peer-to-Peer Network and the network status are classified according to their performances. Based on the concepts of amplifying and shrinking, a self-adaptive schedule algorithm based on the nodes performance and a data-driven algorithm based on buffer-ratio under the TBMM network topology are designed to realize the dynamic maintenance of network topology according to the difference of node performance. That makes it possible that the nodes can be self-scheduled at a low cost of network workload and computation. It realizes highly efficient data request and send algorithm and assures of data distribution with little delay. Based on these theories, this paper gives an overall plan for the streaming media distribution system and realizes its modules and interfaces with C/C++.