QoE-Based Scheduling for Mobile Cloud Services via Stochastic Learning

In this paper, a quality-of-experience (QoE)-based user scheduling scheme for delay-sensitive mobile cloud services (MCS) is proposed. The proposed scheme aims at optimizing the user QoE, which is mainly determined by both the application-level and network-level quality of services. Packet delay, as an essential factor affecting QoE, is discussed under the context of QoE optimization. The optimization problem is modeled as an infinite- horizon average cost Markov Decision Process (MDP), based on both the dynamics of channel state information (CSI) and queue state information (QSI). In order to reduce the exponential memory requirement and computational complexity, a distributed stochastic learning algorithm which only requires local CSI and QSI is introduced. Simulation results show that the proposed scheme can achieve significant improvement in QoE over conventional schemes.