Efficiency of QoE-driven network management in adaptive streaming over HTTP

HTTP adaptive streaming (HAS) technology has been widely implemented in entertainment industries. It allows users to smoothly access representations of content when the network work conditions frequently fluctuate. This mechanism not only improves the perceived quality of user but also benefits the network resource utilization. However, the frequent adaption of bit rate may cause the instability of Quality of Experience (QoE) to premium users who are willing to pay for high and stable perceived quality. Therefore, recently an appropriate network management scheme has been explored in order to control streaming behaviors with respect to the requirements of various types of users. In our previous study, a machine learning based network management system has been proposed as a relevant approach to manage QoE of HAS. In this paper, the performance of the proposed system will be clarified in dealing with a practical problem of bandwidth competition between a HAS player and other application clients.