Torrents on twitter: Explore long-term social relationships in peer-to-peer systems

Haiyang Wang, Feng Wang, Jiangchuan Liu, Ke Xu, Di Wu

Research output: Contribution to journalArticle

4 Scopus citations

Abstract

Peer-to-peer file sharing systems, most notably Bit- Torrent (BT), have achieved tremendous success among Internet users. Recent studies suggest that the long-term relationships among BT peers can be explored to enhance the downloading performance; for example, for re-sharing previously downloaded contents or for effectively collaborating among the peers. However, whether such relationships do exist in real world remains unclear. In this paper, we take a first step towards the real-world applicability of peers' long-term relationship through a measurement based study. We find that 95% peers cannot even meet each other again in the BT networks; therefore, most peers can hardly be organized for further cooperation. This result contradicts to the conventional understanding based on the observed daily arrival pattern in peer-to-peer networks. To better understand this, we revisit the arrival of BT peers as well as their longrange dependence. We find that the peers' arrival patterns are highly diverse; only a limited number of stable peers have clear self-similar and periodic daily arrivals patterns. The arrivals of most peers are, however, quite random with little evidence of long-range dependence. To better utilize these stable peers, we start to explore peers' long-term relationships in specific swarms instead of conventional BT networks. Fortunately, we find that the peers in Twitter-initialized torrents have stronger temporal locality, thus offering great opportunity for improving their degree of sharing. Our PlanetLab experiments further indicate that the incorporation of social relations remarkably accelerates the download completion time. The improvement remains noticeable even in a hybrid system with a small set of social friends only.

Original languageEnglish (US)
Article number6313582
Pages (from-to)95-104
Number of pages10
JournalIEEE Transactions on Network and Service Management
Volume10
Issue number1
DOIs
StatePublished - Mar 2013

Keywords

  • BitTorrent
  • Long-term relationship
  • Self-similar
  • Social networks

Fingerprint Dive into the research topics of 'Torrents on twitter: Explore long-term social relationships in peer-to-peer systems'. Together they form a unique fingerprint.

  • Cite this