Networked Systems Seminar

Seminar #9: Friday, May 29th, 2009
DBH 6011, 11am

Joint talk with the CS Seminar

P2P Live Video Streaming

Keith Ross
Polytechnic Institute of NYU

About the Talk:

With P2P live video streaming, peers viewing videos also assist the server in streaming the videos, thereby significantly reducing server infrastructure cost. In recent years, there have been several large-scale deployments of P2P live video systems; for example, PPstream has 350 million downloads, about 12 million active users every day, and thousands of channels.

These P2P streaming systems have several fundamental performance problems including large channel switching delays, long playback lags, and poor performance for less popular channels. We propose a new cross-channel P2P streaming framework, called View-Upload Decoupling (VUD). VUD strictly decouples peer downloading from uploading, bringing stability to multichannel systems and enabling cross-channel resource sharing. We propose a set of peer assignment and bandwidth allocation algorithms to properly provision bandwidth among channels, and introduce sub-stream swarming to reduce the bandwidth overhead. We evaluate the performance of VUD via analytical models, simulations and a PlanetLab implementation.

The current designs also lack incentives for users to contribute bandwidth resources. To address incentives, we propose, prototype, deploy and validate LayerP2P, a P2P live streaming system with built-in incentives. LayerP2P combines layered video and a tit-for-tat-like algorithm so that a peer contributing more upload bandwidth receives more layers and consequently better video quality. We fully implement LayerP2P (including seeds, clients, trackers, and layered codecs), deploy the prototype in PlanetLab, and perform extensive experiments. Compared to a P2P live streaming system with single-layer video, LayerP2P provides significantly improved video quality for cooperative peers, and free-riders do not adversely affect system performance.

This talk is based on two different papers. The first paper ("Queuing Network Models for Multi-Channel Live Streaming Systems") recently won "Best Paper Award" at INFOCOM 2009.


About the Speaker:

Professor Ross is the Leonard J. Shustek Chair Professor in Computer Science at Polytechnic Institute of NYU since January 2003. Professor Ross has worked in peer-to-peer networking, Internet measurement, video streaming, Web caching, multi-service loss networks, content distribution networks, network security, voice over IP, optimization, queuing theory, and Markov decision processes. He is an IEEE Fellow, associate editor for IEEE/ACM Transactions on Networking, has served as PC chair for several networking and multimedia conferences. He has also served as an advisor to the Federal Trade Commission on P2P file sharing. Professor Ross is co-author (with James F. Kurose) of the popular textbook, Computer Networking: A Top-Down Approach Featuring the Internet, published by Addison-Wesley (first edition in 2000, fifth edition 2009). It is the most popular textbook on computer networks in CS departments, both nationally and internationally; it has been translated into twelve languages. Professor Ross is also the author of the research monograph, Multiservice Loss Models for Broadband Communication Networks, published by Springer in 1995.

If you would like to meet with the speaker, please contact Athina Markopoulou at athina-at-uci-dot-edu.