On the synthetic dataset generation for IPTV services based on user behavior

Abstract

The adoption of the paradigm shift from push-based media broadcasting to pull-based media streaming has seen a significant growth in the recent decade. IPTV is good example to illustrate this claim. In IPTV systems hundreds (or maybe thousands in near future) of live TV channels and video contents are available to subscribers. In many application domains a clear understanding of access pattern to the items is necessary. However, for security reasons, in IPTV systems this kind of information is not publicly available. In this paper, taking into account a model which mimics the behavior of a typical IPTV user, and with the aid of MovieLens dataset, we produce a trace file or synthetic dataset, named UBSDI. We then show that, this dataset can reflect many properties of real datasets quite realistically. This dataset is publically available and can be used in many applications such as recommender systems, network capacity planning, network dimensioning, and system performance optimization.