A holistic modeling for QoE estimation in live video streaming applications over LTE Advanced technologies with Full and Non Reference approaches

Current mobile networks are providing high speed access to Internet at a rate of Gigabits per second. In this scenario, traditional services over wired networks are an alternative, in particular those based on live video streaming. But in the transition, different issues should be considered due to the rapid changing network conditions and the limited resources of the mobile devices. These issues should be taken into account to keep a good Quality of Experience (QoE) of the video in terms of a high Mean Opinion Score (MOS), a subjective video quality. Our goal is to estimate and predict this subjective metric in a holistic manner. Thus, we have analyzed and measured different variables related to Quality of Service, bit stream and basic video quality metrics of live video streaming over LTE Advanced (4G) mobile networks, as well as server side. With these variables, we apply a statistical method, called Factor Analysis based on the correlation of the variables, which allows us to estimate MOS of the delivered video in a robust and reliable way, both in Full Reference and Non Reference approaches. The real MOS has been evaluated through surveys. Finally, we compare the accuracy of the estimated MOS for the different approaches and compare their performance with well known publicly available video quality algorithms.