Authors: Takaya Fujimoto; Rei Endo; Hiroshi Shigeno
Addresses: Graduate School of Science and Technology, Keio University, 3-14-1 Hiyoshi Kouhokuku, Yokohama Kanagawa, 223-8522, Japan. ' Graduate School of Science and Technology, Keio University, 3-14-1 Hiyoshi Kouhokuku, Yokohama Kanagawa, 223-8522, Japan. ' Faculty of Science and Technology, Keio University, 3-14-1 Hiyoshi Kouhokuku, Yokohama Kanagawa, 223-8522, Japan
Abstract: In peer-to-peer video-on-demand streaming, peers reduce video server load by caching viewed video data and sending these data to other peers. Although a first-in, first-out approach is typically used in the caching algorithm, it is not an efficient use of peers' upload capacity because the peers can cache the data of unpopular videos that no other peers request. In this paper, we propose a video-popularity-based caching and reservation (VPCR) scheme that increases the utilisation of peers' upload capacity. Through caching based on video popularity, even when peers view unpopular videos, they can deliver popular videos to other peers. The reservation scheme addresses temporal viewership fluctuations for each video which decrease utilisation of peers' upload capacity in VPCR. We evaluate VPCR through computer simulations and show that it is efficient in reducing the video server load under various video popularity distributions. We also show the relation between video server load and the number of simultaneous reservation acceptances for each peer.
Keywords: P2P; peer-to-peer; VoD streaming; video-on-demand; video popularity; load distribution; temporal fluctuation; caching; reservation; popularity distributions; viewer fluctuations; upload capacity; simulation; server load.
International Journal of Grid and Utility Computing, 2012 Vol.3 No.2/3, pp.188 - 199
Received: 11 Jun 2011
Accepted: 06 Dec 2011
Published online: 20 Dec 2014 *