A Scheme for Maximal Resource Utilization in Peer-to-Peer Live Streaming

TL;DR

This paper introduces a practical scheduling algorithm for peer-to-peer live streaming that maximizes resource utilization and streaming rate, even under conditions with no spare capacity, by optimizing overlay network structure and sub-stream distribution.

Contribution

It presents a novel scheduling scheme that efficiently utilizes peer and server capacities, constructs short delay trees, and outperforms existing methods like SplitStream in simulations.

Findings

Peers dynamically converge to efficient overlay structures

The scheme maximizes streaming rate under capacity constraints

Short hop-count delay trees are achieved

Abstract

Peer-to-Peer streaming technology has become one of the major Internet applications as it offers the opportunity of broadcasting high quality video content to a large number of peers with low costs. It is widely accepted that with the efficient utilization of peers and server's upload capacities, peers can enjoy watching a high bit rate video with minimal end-to-end delay. In this paper, we present a practical scheduling algorithm that works in the challenging condition where no spare capacity is available, i.e., it maximally utilizes the resources and broadcasts the maximum streaming rate. Each peer contacts with only a small number of neighbours in the overlay network and autonomously subscribes to sub-streams according to a budget-model in such a way that the number of peers forwarding exactly one sub-stream will be maximized. The hop-count delay is also taken into account to construct a short depth trees. Finally, we show through simulation that peers dynamically converge to an efficient overlay structure with a short hop-count delay. Moreover, the proposed scheme gives nice features in the homogeneous case and overcomes SplitStream in all simulated scenarios.