Modeling Video on Demand services taking into account statistical dependences in user behavior

The popularity of video services on the Internet, and moreover High Definition (HD) videos, has increased continuously in recent years. This growing demand for high quality video services can cause problems in current communication networks. For this reason, the characterization and modeling of video workloads have become essential factors in evaluating the performance of these services. This paper presents the implementation of a simulation model of a Video on Demand (VoD) service. The model takes into account both the behavior of the users and the server workload. While the former is analyzed through a real VoD service, the latter is analyzed through lab experiments. Nevertheless, the main contribution of our model is that it also considers the dependences among user interactions. The model has been validated using real data and evaluated in different situations. Results show that performance is influenced both by massive access to the contents and by the dependence structure of user interactions.