Resource Management In Mobile Cloud Computing Networks For User QoS And QoE

In Mobile cloud computing(MCC) system, users receive the cloud computing services through mobile networks, both cloud and wireless resources could be the bottleneck for performance improvement. The current independent management for these two kinds of resources in MCC system cannot guarantee the end-to-end user quality-of-service(QoS) and quality-of-experience(QoE). Considering the problems that the heterogeneity of user requirements, mobile networks and cloud systems has been ignored in MCC, this thesis mainly studies the resource management in MCC networks for user QoS and QoE, including restless bandits based joint cloud and radio resource management for video transmissions in mobile cloud computing networks, QoS-aware dynamic resource management in heterogeneous mobile cloud computing networks and Stackelberg Game based green mobile cloud game system.This thesis studies the joint cloud and radio resource management in MCC networks for video transmission in the first place. The joint resource management could improve the performance of MCC system by optimizing the important video application layer parameter, intra-refreshing rate. The Application Layer video QoS parameter performance, distortion, is adopted as the performance metrics. We formulate this problem as a stochastic restless bandits optimization problem, which has the ?indexibility? feature and the optimization problem can be transformed into a distributed selection problem by selecting the priority-index. In the off-line stage, the priority-index could be obtained by linear programming relaxation and primal-dual index heuristic which could be stored in the index tables. In the on-line stage, what the users need to do is to lookup the tables to decide the optimal index according to the current state, which guarantees the low computational complexity.Besides, this thesis studies the QoS-aware dynamic resource management in heterogeneous MCC networks, which aims to design the network as the integration of the mobile access part and the cloud computing part, utilizing the inherent heterogeneity to meet the diverse QoS requirements of tenants. And then a QoS-aware novel cross-network radio and cloud resource management scheme has been proposed. The proposed scheme is also formulated as a Restless Bandits problem.Furthermore, a Stackelberg Game based cloud gaming system for cloud game business is proposed in this thesis. In this system, the design object is the novel mobile cloud game system framework to enhance the interactivity of users and take resource allocation problem of cloud servers and users into account, including the quantity and the price of onloading components, and the power consumption for transmission. We formulate the resource management problem as three-stage Stackelberg model. The backward induction method is used to solve the utility function. The architecture of game system and dynamic interaction process are conducted for reasonable resource allocation, balancing the relationship among cloud, mobile network and user terminal and making game system greener and more efficient while satisfying the QoS and QoE in multi-user environment.