Live and On-Demand Streaming Systems with Network Coding
Since 2007, we are interested in the application of network coding in large-scale streaming systems.
R2: Practical peer-to-peer streaming with network coding: In this work, we are interested in the design, analysis, and implementation of a new peer-to-peer streaming protocol that takes full advantage of network coding, and hopefully combines the benefits of tree-based per-slice push and mesh-based per-segment pull, with respect to a small number of important performance metrics in peer-to-peer streaming.
A theoretical framework on large-scale P2P live streaming systems with network coding: In this work, we have developed a theoretical framework to help us understand how network coding has simplified the design of P2P live streaming systems, with respect to playback quality, initial buffering delay, server bandwidth costs, as well as the resilience to flash crowd and severe peer dynamics.
UUSee: large-scale operational on-demand streaming with network coding: In this work, we wish to present the objectives, rationale, and design in the first production deployment of random network coding. Before this piece of work, network coding has not been deployed in real-world commercial systems in operation at a large scale, and in a production setting. This work describes how network coding can be used as the cornerstone of a large-scale production on-demand streaming system, operated by UUSee Inc., delivering thousands of on-demand video channels to millions of unique visitors each month. To achieve a thorough understanding of the performance of network coding, we have collected 200 Gigabytes worth of real-world traces throughout the 17-day Summer Olympic Games in August 2008, and present our lessons learned after an in-depth trace-driven analysis.
