Using K-Hop Rarest First Policy For Piece Selection And Request Selection To Reduce Download Time In BitTorrent

碩士 === 輔仁大學 === 資訊工程學系 === 100 === Traditional client-server architecture has become inadequate in face of today’s growing demand of file sharing due to the bandwidth bottleneck and single point failure problems. An effective solution to the problems is to use Peer-to-peer (P2P) architecture, among...

Full description

Bibliographic Details
Main Authors: Hsu, Nien-En, 徐念恩
Other Authors: Lu, Chun-Hsien
Format: Others
Language:zh-TW
Published: 2012
Online Access:http://ndltd.ncl.edu.tw/handle/67940724411090622023
Description
Summary:碩士 === 輔仁大學 === 資訊工程學系 === 100 === Traditional client-server architecture has become inadequate in face of today’s growing demand of file sharing due to the bandwidth bottleneck and single point failure problems. An effective solution to the problems is to use Peer-to-peer (P2P) architecture, among which the BitTorrent (BT) protocol has been most widely used. Because the tracker does not consider the piece distribution and the structure of topology when it sends the peer list to each joining peer, the swarm may not form a topology that facilitates the quick spread of difference pieces. In order to reduce the average file download time, it is important to have good strategies on both piece selection and request selection. In this paper, we proposed a k-hop rarest first (KHRF) strategy to replace the local rarest first and tit-for-tat strategies in BT. Under this strategy, a peer will choose to first download piece that is rarest within k hops. When deciding which request to serve, a peer will serve the piece request that has the maximum rareness within k hops. Simulation results show that the KHRF strategy can reduce about 21% of the average download time, and 18% of the total elapsed time. KHRF can improve the availability of BT system.