CDN-P2P Streaming Media Transmission System Based On Neighbor Recommendation Design And Impletation

In recent years with the development of the Internet and streaming media technology, streaming media services is becoming an important application of the Internet, and tends to a large-scale, high-quality direction. The Large-scale streaming media service was also known as research hotspot recently.Research shows that combine the advantages of CDN and P2P technology to apply them to streaming media transmission system can significantly improve the performance of streaming media transmission. This paper focuses on the following three aspects:Firstly, for the system of network nodes with characteristics of high degree of autonomy and dynamic, in order to make the nodes in the system can obtain a more stable and smooth media streaming service, we add a network distance and stability based neighbor recommendation function to the system. The nodes in the system can be divided into domains according to the network distance, and to help the nodes to select neighbors which are more stable and small transmission delay. Then nodes with high stability are fully matched which improves the quality of service.Secondly, as the increasing number of nodes and resources, we added a user-based recommendation function to help users to find the resources they might like. But collaborative filtering needs a large amount of calculation on center server. So we improved the original recommendation algorithm, forming a distributed collaborative filtering recommending algorithm based on interest subnet. This algorithm allocates similarity computing tasks to nodes, and builds interest subnet spontaneously. Then complete the recommendation by the central server. The algorithm reduced the amount of computation process by center server, and insured recommended accuracy.Thirdly, in the system, node’s cache is limited. The node can provide a transport service only when it has the cache segments which the requesting nodes require. Obviously, the more nodes provide transport services, the less pressure the server will have. To solve this problem, we added a heat and segment priority based FIFO buffer replacement strategy to improve the number of service providing nodes, reducing the pressure on the server nodes and the delay.