Optimization Study, Based On The Bittorrent Peer To Peer Network Node Selection Algorithm

With the development of information technology, the Internet is gradually been accepted and used by the public. BitTorrent networks based on P2P appear. Servers are no longer needed by the Internet since it inherits the advantages of P2P that every node in the net is both a client and a server. Resources are shared from points to points among customers that make the Internet resources in full use. Searching on BitTorrent networks have significant practical and theoretical meanings in developing P2P networks and the Internet resources.This thesis elaborates the principle of BitTorrent based on analyzing the main function and original code of the BitTorrent networks' structure, protocol, core algorithm, server and client. Tracker servers play an important role in the network communication; furthermore, it decides the performance of the whole BitTorrent networks. The paper especially analyzes the function and algorithm of the Tracker server.The paper analyzes in detail the relevant protocol and major algorithm of Tracker and found that there exist some limitations in the original algorithm. As the result, an optimized plan is put forward so that the downloading speed of the BitTorrent networks nodes can be increased. Based on a great deal of experiments, according to the tracking result during the process of node's downloading to reflect the characteristic of the speed of node's downloading in different stages,and then choose a reasonable threshold value. It deals with the queries in phases on the basis of their downloading speeds, decreasing the percentage of seed nodes at the initial stages and increasing it at the final stages so that the download can be completed in a shorter time and the other nodes can continue to be at service. The rest of the stages follow the original algorithm. In this way, the growth rate of seed nodes in the BitTorrent networks is increased and the download can be speeded up too.Finally, the algorithm constructs 100 downloading nodes in simulative web, generating their original downloading speed by random and let them download a same file at the same time. Then, download a file by the original algorithm and the improved algorithm respectively. By making comparison between the results of two experiments, we can prove that the improved algorithm is better than the original algorithm in terms of downloading speed.