Practical QoE Evaluation of Adaptive Video Streaming

Abstract

Video streaming is an increasingly popular service on the Internet. In HTTP adaptive video streaming (HAS), the video is played while being downloaded, and the quality is selected according to the available bandwidth. Due to this, variations in the transmission affect the playback. The quality of the playback can be rated by technical parameters, which can be grouped by the term ‘Quality of Service’ (QoS), like the video quality, the number and duration of stallings or the time until the video starts playing. These metrics differently influence the user experience. Up to now, no widely accepted model for the Quality of Experience (QoE) for HAS exists. Therefore, we use two conceptually different models and investigate their impact on the resulting QoE. To do so, we use a typical video player, namely the Shaka Player, that can be embedded into websites, and change its buffer configuration. The observed data is then used to evaluate the quality of experience (QoE), combining it into a single ‘Mean opinion score’ (MOS). It can be shown, that, with limitations, these methods can be suited for QoE evaluation.