Four months in daily motion: Dissecting user video requests
Yannick Carlinet, The Dang Huynh, Bruno Kauffmann, Fabien Mathieu, Ludovic Noirie, Sébastien Tixeuil · 2012
The growth of User-Generated Content (UGC) traffic makes the understanding of its nature a priority for network operators, content providers and equipment suppliers. In this paper, we study a four-month dataset that logs all video requests to DailyMotion made by a fixed subset of users. We were able to infer user sessions from raw data, to propose a Markovian model of these sessions, and to study video popularity and its evolution over time. The presented results are a first step for synthesizing an artificial (but realistic) traffic that could be used in simulations or experimental testbeds.