Technology advances have brought a growth in live streaming, where users can broadcast real-time videos of the games they are playing to interested viewers. While characteristics of traditional video on demand sites are relatively well understood, live streaming has received comparatively less attention. A better understanding of such systems is needed to plan for future systems and architectures. This paper uses a crawler and stream analyzer to characterize Twitch.tv, the most popular live streaming system for games. Analysis of the data confirms live stream volume from previous research, while providing novel frame rate and bitrate analysis.

Download:

• Paper: pdf
  
• Slides: pptx
  

See also:

• Dan Farrington and Nick Muesch. Analysis of the Characteristics and Content of Twitch Live-Streaming, Interactive Qualifying Project IQP-MLC-TT14, Worcester Polytechnic Institute, Spring 2015. (Advisor Mark Claypool) Online at: http://www.cs.wpi.edu/~claypool/iqp/twitch-tv/