Our work on user behavior characterization for a mobile live streaming system has been accepted to appear in ACM WWW-web science track, 2015

The accept rate is 13.8%.


Streaming live content to mobile terminals has become prevalent. While there are extensive measurement studies of non-mobile live streaming (and in particular P2P live streaming) and video-on-demand (both mobile and non-mobile), user behavior in \emph{mobile live streaming} systems is yet to be explored. This paper relies on over 4 million access logs collected from the PPTV live streaming system to study the viewing behavior and user activity pattern, with emphasis on the discrepancies that might exist when users access the live streaming system catalog from mobile and non-mobile terminals. We observe high rates of abandoned viewing sessions for mobile users and identify different reasons of that behavior for 3G- and WiFi-based views. We further examine the structure of abandoned sessions due to connection performance issues from the perspectives of time of day and mobile device types. To understand the user pattern, we analyze user activity distribution, user geographical distribution as well as user arrival/departure rates.


Zhenyu Li, Mohamed-Ali Kaafar, Kave Salamatian, Gaogang Xie, User Behavior Characterization of a Large-scale Mobile Live Streaming System, ACM WWW-web science track, 2015 PDF