1. An Empirical Study of Flash Crowd Dynamics in a P2P-based Live Video Streaming System Bo Li, Gabriel Y. Keung, Susu Xie, Fangming Liu , Ye Sun, and Hao Yin Email: [email_address] Hong Kong University of Science & Technology Dec 2, 2008 @ IEEE GLOBECOM, New Orleans
Peer-assisted live video streaming is another attractive service in the Internet: 1) Take PPLive, one of the most popular peer-assisted streaming systems nowadays, as an example: at the end of 2005, it has 20 million download and 1 million; independent viewers per day [2]. According to [4], it supports over 200 ; 000 concurrent users at bit rate in the 400-800 Kbps for 2006 Spring Festival Gala on Chinese New Year on January 28, 2006. In 2007, the number of concurrent users for the most popular PPLive session raises to 1 : 5 million [3]. This corresponds to an aggregate bit rate in the vicinity of 600 Gbps, or 540 TB transferred per the 2 hour event. Actually, this type of service but with more challenges than that of file sharing: First,
, either peers joining process is stretched over a longer period of time or application itself can tolerate much longer delay , the dynamic of user population during annual Spring Festival Gala on Chinese New Year
In face of frequent peer churns, the maintenance of streaming tree(s) in tree-based approaches is still challenging, and the recovery of tree(s) incurs extra cost. Recently, mesh-based approaches (also referred to as data-driven approaches) have been adopted in many large-scale peer-assisted live video streaming systems, such as Coolstreaming [32], PPLive, UUSee, and etc. In contrast with tree-based approaches, mesh-based overlay designs do not construct and maintain an explicit structure for delivering data. 4) Generally, a mesh-based streaming system has a tracker to keep track of peers in the video session. 5) A peer may download/upload segments from/to multiple partners simultaneously
over 20% of the users have tried 1 or 2 times in order to successfully start a video session. Hence, a flash crowd has significant impact on the initial joining phase in a P2P streaming system.
it takes longer time for a newly joined peer to obtain video stream
Besides the above, there are also open issues relevant to multiple channels, ISPs which is covered in my report, but not shown here due to time limit