Providing Video-on-Demand using Peer-to-Peer Networks

TLDR

Peer‑to‑peer networks provide fast, scalable media delivery, yet users typically must download the entire video before playback, and little research has tackled low start‑up latency and sustained playback for near‑Video‑on‑Demand. The authors aim to show that high‑quality near‑Video‑on‑Demand can be achieved using P2P swarming systems. They study block scheduling in a mesh‑based P2P network, demonstrating that pre‑fetching and network coding significantly improve data dissemination. Experiments on 120‑minute videos reveal that a 10‑minute buffer (≈8% of the video) enables playback rates up to 80–90% of the access link capacity, even with dynamic user population changes.

Abstract

Digital media companies have recently started embracing peer-assisted distribution networks as an alternative to traditional client-server architectures [5]. Such Peer-to-Peer (P2P) architectures ensure a fast and scalable delivery of media content. However, their drawback is that users need to often wait for the full video to be downloaded before they can start watching it. While a lot of effort has gone into optimizing the distribution of large video files using P2P swarming techniques, little research has been done on how to ensure a small start-up time and a sustainable playback rate to enable a play-asyou-download experience. In this work, we address the challenges underlying the problem of near Video-on-Demand (nVoD) using P2P swarming systems, and provide evidence that high-quality nVoD is feasible. In particular, we investigate the scheduling problem of efficiently disseminating the blocks of a video file in a P2P mesh-based system, and show that pre-fetching and network coding techniques can provide significant benefits. Our results show that, for videos which are 120 minutes in duration, 10 minutes (≈ 8% of the video’s length) of buffering at start-up can enable playback rates that are close (up to 80 − 90%) to the access link capacity, even under dynamic changes of the user population.