BitTorrent is the most popular file sharing protocol on the Internet. It is proved that its performance are near-optimal for generic file distribution when the overlay is in a steady state. The two main BitTorrent strategies, tit-for-tat and local rarest fist, work at best: the former assures reciprocity in downloading and uploading rates between peers and the latter distributes the different file pieces equally among the overlay. Both assure good performance in terms of resource utilization and the practical consequence is that the peers achieve good downloading times. The best network condition for the protocol is a network characterized by roughly fixed arrival rates and no flash crowds. Nevertheless, many research works argue that the performance of the protocol quickly degrades when the peers join and leave at high rates, the network is affected by flash crowds phenomenon and the number of peer that shares the complete file is only a little fraction of the total population. This is the case of many real-world peer-to-peer applications like video-on-demand or live-streaming. In this scenario the introduction of some kind of network coding can mitigate the adverse network behavior and improve the overall performance of the protocol. In this paper we develop a modification of the BitTorrent protocol with the introduction of Luby-Transform codes, that belong to the class of the erasure rateless codes. Using a modified version of GPS (General Purpouse Simulator), we set up simulations that prove how these changes make the original protocol more robust to adverse network conditions and speed up its performance in such situations.

BitTorrent and fountain codes: friends or foes?

SPOTO, SALVATORE;GAETA, Rossano;GRANGETTO, Marco;SERENO, Matteo
2010

Abstract

BitTorrent is the most popular file sharing protocol on the Internet. It is proved that its performance are near-optimal for generic file distribution when the overlay is in a steady state. The two main BitTorrent strategies, tit-for-tat and local rarest fist, work at best: the former assures reciprocity in downloading and uploading rates between peers and the latter distributes the different file pieces equally among the overlay. Both assure good performance in terms of resource utilization and the practical consequence is that the peers achieve good downloading times. The best network condition for the protocol is a network characterized by roughly fixed arrival rates and no flash crowds. Nevertheless, many research works argue that the performance of the protocol quickly degrades when the peers join and leave at high rates, the network is affected by flash crowds phenomenon and the number of peer that shares the complete file is only a little fraction of the total population. This is the case of many real-world peer-to-peer applications like video-on-demand or live-streaming. In this scenario the introduction of some kind of network coding can mitigate the adverse network behavior and improve the overall performance of the protocol. In this paper we develop a modification of the BitTorrent protocol with the introduction of Luby-Transform codes, that belong to the class of the erasure rateless codes. Using a modified version of GPS (General Purpouse Simulator), we set up simulations that prove how these changes make the original protocol more robust to adverse network conditions and speed up its performance in such situations.
7th International Workshop on Hot Topics in Peer-to-Peer Systems (HOT-P2P), in conjunction with IPDPS 2010
Atlanta, GE, USA
19-23 April 2010
Proceedings of the 7th International Workshop on Hot Topics in Peer-to-Peer Systems (HOT-P2P)
IEEE Computer Society Press
1
8
9781424465330
Spoto S.; Gaeta R. ; Grangetto M. ; Sereno M.
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/2318/117056
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 8
  • ???jsp.display-item.citation.isi??? ND
social impact