Title Closest playback-point first: A new peer selection algorithm for P2P VoD systems

Z Wen, N Liu, Kl Yeung, Lei
Peer-to-peer (P2P) based video-on-demand (VoD) streaming service has been gaining popularity recently. Unlike live streaming, a VoD peer always starts its playback from the beginning of a stored video. The playback-points of different peers, as well as the amount of video contents/pieces they cached, depend on when they join the video session, or their viewing ages. As a result, the upload bandwidth of younger peers tends to be underutilized because older peers are not interested in their
more » ... sted in their cached video pieces. The collaborative piece exchange among peers is undermined due to the unbalanced supply and demand. To address this issue, a playback-point based request peer selection algorithm is proposed in this paper. Specifically, when a peer requests a particular video piece, among the set of potential providers, a request is sent to the peer that has the smallest playback-point difference with itself. We call this request peer selection algorithm closest playback-point first (CPF). With CPF, peers with similar available content can be loosely grouped together for a more balanced collaborative piece exchange. Extensive packet-level simulations show that with CPF, the video playback quality is enhanced and the VoD server load is significantly reduced. I.