Reducing Average Download Time in BitTorrent-like Systems via Strategic and Dynamic Topology Adjustment

碩士 === 輔仁大學 === 資訊工程學系碩士班 === 104 === Because of the explosive growth of network usage, traditional Client-Server architecture has been replaced by Peer-to-peer (P2P) architecture. BitTorrent has been the most popular P2P file sharing system, in which a peer upon joining the system will be given a c...

Full description

Bibliographic Details
Main Authors: HONG,YI-SYUAN, 洪翊軒
Other Authors: LU.CHUN-HSIEN
Format: Others
Language:zh-TW
Published: 2016
Online Access:http://ndltd.ncl.edu.tw/handle/32409327084495937522
Description
Summary:碩士 === 輔仁大學 === 資訊工程學系碩士班 === 104 === Because of the explosive growth of network usage, traditional Client-Server architecture has been replaced by Peer-to-peer (P2P) architecture. BitTorrent has been the most popular P2P file sharing system, in which a peer upon joining the system will be given a candidate neighbor list to set up neighboring connections, and those connections will not change till the end of file downloading. This situation brings some problems. For example, if none of the neighbors has the piece that a peer lacks, that a peer will wait until its neighbors acquires the piece, also if the high-bandwidth and low-bandwidth peers are allocated in the neighbors unevenly, high-bandwidth peers may quickly complete file downloading and leave the system, while low-bandwidth peers are left behind to download the pieces slowly in the system. We propose an algorithm called Strategic and Dynamic Topology Adjustment (SDTA) to reduce average download time in Bittorrent-like system. When a peer is downloading inefficiently, it will try to add a new neighbor based on its own degree of completion and uploading bandwidth. Our goal is to let high-bandwidth peers reach a high completion state, so that they become helping peers to serve low-bandwidth peers. In addition, adding a new neighbor can help reduce waiting time when a peer cannot find a desirable piece among its neighbors. Simulation results show that our SDTA algorithm can produce a thirteen to twenty-eight percent reduction over BT in the average download time.