Research Article
Reducing cross-network traffic in P2P systems via localized neighbor selection
@INPROCEEDINGS{10.1109/CHINACOM.2009.5339819, author={Lijie Sheng and Haoyu Wen}, title={Reducing cross-network traffic in P2P systems via localized neighbor selection}, proceedings={4th International ICST Conference on Communications and Networking in China}, publisher={IEEE}, proceedings_a={CHINACOM}, year={2009}, month={11}, keywords={peer to peer; P2P; BitTorrent; overlay network; neighbor selection; SRF}, doi={10.1109/CHINACOM.2009.5339819} }- Lijie Sheng
Haoyu Wen
Year: 2009
Reducing cross-network traffic in P2P systems via localized neighbor selection
CHINACOM
IEEE
DOI: 10.1109/CHINACOM.2009.5339819
Abstract
P2P software bring convenient and pleasure to P2P users, but become a problem of ISP, because they produce huge traffic on network, especially on cross-network links, which increases financial burden on ISP. The reason partially comes from the fact that when P2P software build their overlay network, they ignore the underlying network topology, and select neighbors randomly, so most of the traffic is between remote peers and crosses several networks. To overcome that traffic explosion problem, we propose a simple mechanism named SRF, in which most of the connections are kept locally between nearby peers, but a few shortcuts between remote peers can make P2P network robust and efficient to propagate files or media streams rapidly. With analysis we demonstrate that the overlay network constructed by SRF is ldquosmall worldrdquo network, while the overlay network constructed by regular BitTorrent is ldquorandomrdquo network. Simulation results show that SRF can reduce cross-network traffic remarkably and can accelerate download explicitly at the same time. Thus SRF benefits both ISP and P2P users.
