PONDER: Performance Aware P2P Video-on-Demand Service

Abstract

Providing scalable video-on-demand (VoD) service over the Internet is desirable however technically challenging due to stringent resource and performance requirement. In this paper, we propose PONDER, a performance aware Peer-to-peer video- <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">ON</i> - <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">DE</i> mand se <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</i> vice that is capable of providing high quality VoD yet is much more scalable than traditional server-client service model. PONDER employs a dual approach that incorporates the mesh-based p2p downloading into the server-client VoD service. The p2p downloading serves a significant portion of data thereby reducing the load on the server. Meanwhile, the server devotes its resources to offering viewers instantaneous playback and urgent data to meet the performance requirement. A set of devices is developed to make the design feasible in practice. Simulation results show that PONDER increases the number of admitted clients by a factor of 15 when the average uplink bandwidth is roughly equal to the playback rate; and users' viewing quality is comparable to the one offered by unicast with tiny fraction of video data missing the playback time.