SDN-based in-network early QoE prediction for stable route selection on multi-path network

Abstract

As QoE is useful to uniformly handle many kinds of application flows, we have been tackling QoE-oriented network resource management based on SDN technology. Toward this goal, our previous study proposed a QoE measurement method for on-going streaming flows. However, the standard QoE calculation model requires at least 8 s for collecting the flow information. In this study, we tackle early QoE prediction on a SDN-enabled multi-path network. To predict video QoE as soon as possible, we exploit not only packet loss rate measured regularly but also the number of packet transmissions by short-period measurement at the flow arrival. Finally, through experiments, we demonstrated that QoE of all flows can be maximized by selecting an appropriate route based on the predicted QoE.