On Modeling and QoE Evaluation of Buffered Video Streaming in Multi-Cellular Networks

Future mobile wireless networks are expected to cope with an increasing demand of various applications, in which video streaming is involved. These applications are not only limited to the traditional (enhanced) mobile broadband use case, but comprise also use cases, such as augmented reality, that include video delivery and require low latency and uninterrupted playback. Therefore, it is important to understand the video performance of multiple end-users in a multi-cellular network. In this paper, we propose mathematical models, to derive video performance indicators, namely the startup delay distribution and the buffer starvation probability. In addition, we introduce solutions to approximate and numerically solve the underlying systems of ordinary and partial differential equations of the proposed models. Furthermore, a quality of experience metric is introduced that maps the video performance to user satisfaction. This metric can be taken as a target function for network optimization.