An Application-aware Scheduling Policy for Real-time Traffic

The pervasiveness of mobile applications stimulates more eager demand for Quality of Experience (QoE) than Quality of Service (QoS), especially on the aspect of link scheduling in wireless networks. In many applications, end users concern more about transmission quality of an individual task rather than an individual packet. A task may correspond to a piece of video, music, etc. and may include many packets. This paper proposes a new network model aiming at improving users' experience that pushes the scheduling problem to the task layer. We first introduce a QoE requirement that can generalize the QoS requirement in link scheduling, the partial result requirement. Subsequently, a novel scheduling policy is proposed which can capture this requirement for each task, and then performs an application-aware scheduling. We theoretically analyze the performance of the novel scheduling policy, and discuss the impact of the QoE requirements and network settings. Finally, the simulation results indicate that our scheduling policy can significantly improve QoE.