Chunk Distribution in Mesh-Based Large-Scale P2P Streaming Systems: A Fluid Approach

We consider large-scale mesh-based P2P systems for the distribution of real-time video content. Our goal is to study the impact that different design choices adopted while building the overlay topology may have on the system performance. In particular, we show that the adoption of different strategies leads to overlay topologies with different macroscopic properties. Representing the possible overlay topologies with different families of random graphs, we develop simple, yet accurate, fluid models that capture the dominant dynamics of the chunk distribution process over several families of random graphs. Our fluid models allow us to compare the performance of different strategies providing a guidance for the design of new and more efficient systems. In particular, we show that system performance can be significantly improved when possibly available information about peers location and/or peer access bandwidth is carefully exploited in the overlay topology formation process.

1 Introduction

Recently, we have been witnessing the emergence of a new class of popular P2P applications, namely, large-scale P2P video streaming. Both live and on-demand P2P streaming systems have the potential of changing the way we watch TV, providing ubiquitous access to a vast number of channels, personalizing the user's experience, and enabling roaming services.