Characterization and Modelling of YouTube Traffic in Mobile Networks

Video streaming is one of the most data-intensive applications of today’s Mobile Internet and YouTube generates 20% of mobile networks downstream traffic in several regions. YouTube employs the progressive download technique for video playback and therefore its traffic is bursty. We present the characteristics of the most important burst measures of traffic generated by YouTube when accessed via mobile broadband connections. Moreover, we also distinguish the characterization for non-optimized and optimized traffic since mobile operators are using media optimization platforms to effectively deliver video content. In this paper, we present our measurement-based analytical results to derive characterization of the traffic sources. As a result, we propose a generic model and its parameters according to the optimized and non-optimized traffic sources based on the experimental evaluation of the captured YouTube traffic. The proposed traffic models can be used in simulation of future work. Keywords-YouTube; video optimization; burstiness; traffic characterization; traffic model.