Network Coder Placement for Peer-to-Peer Content Distribution

Dinh NGUYEN  Hidenori NAKAZATO  

IEICE TRANSACTIONS on Communications   Vol.E96-B   No.7   pp.1661-1669
Publication Date: 2013/07/01
Online ISSN: 1745-1345
DOI: 10.1587/transcom.E96.B.1661
Print ISSN: 0916-8516
Type of Manuscript: Special Section PAPER (Special Section on Internet Architectures, Protocols, and Management Methods that Enable Sustainable Development)
peer-to-peer,  content distribution,  network coding,  

Full Text: PDF(1.1MB)>>
Buy this Article

We study the use of network coding to speed up content distribution in peer-to-peer (P2P) networks. Our goal is to get the underlying reason for network coding's improved performance in P2P content distribution and to optimize resource consumption of network coding. We observe analytically and experimentally that in pure P2P networks, a considerable amount of data is sent multiple times from one peer to another when there are multiple paths connecting those two particular peers. Network coding, on the other hand, when applied at upstream peers, eliminates information duplication on paths to downstream peers, which results in more efficient content distribution. Based on that insight, we propose a network coder placement algorithm which achieves comparable distribution time as network coding, yet substantially reduces the number of encoders compared to a pure network coding solution in which all peers have to encode. Our placement method puts encoders at critical network positions to eliminate information duplication the most, thus, effectively shortens distribution time with just a portion of encoders.