A popularity-based neighbor selection model in P2P file-sharing system

In the past few years, P2P file distribution applications (e.g., Bit Torrent) are becoming so popular that they are the dominating source of Internet traffic. This brings significant problems not only because the increasing P2P traffic costs most of network bandwidth, but also the overmuch P2P connections increase the burden on the end-users. In this paper we present a peers' popularity model that rank these peers by performance and then select the strong peers, those who can provide efficient uploading and who have sufficient load capacity. To achieve this goal, we design a general definition to peers' popularity and analyze it with a Zipf-like distribution. Furthermore, we carry out the implementation proposals in most P2P systems with tracker server. Through this model we can reduce the superfluous P2P connections and increase the efficiency of P2P download.