On Algorithms Of Content Distribution Service In P2P Enviroment

Content distribution service's goal is to distribute the same file to multiple users in the network, which has always been the key subject of Internet study, and the main problem of it is how to achieve the maximum usage of network bandwidth. As the Internet is developing rapidly, the growing number of users overwhelmed the traditional C/S architecture, while the IP Multicast could not be readily implemented in the wide area network, which forced people to search better solutions of content distribution in the P2P environment of application layer.This thesis has the following contributions:1. We classified the P2P environment into two scenarios, which are cooperative and incooperative, summarized the recent related works in the field of content distribution in the P2P environment and divided their algorithms into structured or unstructured and encoded or non-encoded four groups. Finally we discuss all kinds of typical algorithms in detail.2. for single multicast tree construction of the typical structured algorithm can't fully utilize the upload bandwidth of leaf nodes, we proposed a gradual way of construct multicast trees - U tree packing algorithm. Our object is to optimize the bandwidth usage in polynomial time. Finally, we made some simulation experiment to observe the performance under the Internet topology generated by INET and compared the throughput of U tree packing with the traditional tree-built structured algorithms.