面向下一代移动应用的移动云服务关键技术研究

软件定义网络（SDN）与网络功能虚拟化（NFV）是一种软件集中控制、硬件虚拟化管理、资源可动态分配和拓扑可灵活重构的网络优化技术框架，是实现下一代移动云服务的关键技术。本项目针对大规模移动云服务网络体系架构，将SDN/NFV与传统移动网络有机融合，研究网络控制与数据转发的分离机制，实现移动网络拓扑的快速响应与自适应扩展；将云计算引入接入网络，研究移动服务请求与云接入资源的映射关系，提出在线滚动优化的分布式资源共享策略，分析大规模移动接入的分布特性，提出基于联盟形成博弈的协作传输机制，建立频谱接入的动态优化模型；针对异构场景下移动设备的计算密集型任务的随机特性，建立云辅助的移动任务卸载模型，设计卸载可接受的服务质量约束，提出基于在线学习机制的动态任务卸决策算法。本项目尝试为现有移动云服务网络设计一种新的可扩展架构和一系列服务优化框架，提高大规模接入移动云服务网络的可扩展性与服务质量。

Software Defined Networking (SDN) and Network Function Virtualization (NFV) technologies enable logically centralized software control, hardware virtualization management, dynamic resource allocation and flexible topology refactoring, which have been treated as critical network optimization frameworks for the next-generation mobile cloud services. For large-scale mobile cloud services, the project introduces SDN/NFV and hierarchical mechanisms into traditional mobile networks, and investigates the separation of the control plane from the data plane, realizing the rapid response and adaptive extension of the mobile network topology. Then in the cloud radio access network, based on the online rolling horizon optimization, the mapping between mobile service requests and cloud radio resources is investigated with the distributed resource sharing strategy. By analyzing the distributed characteristics of large-scale mobile access services, the collaborative data transmission mechanism is studied based on the coalition formation game theory, and the dynamic spectrum access optimization model is formulated. At last, due to the stochastic characteristics of computation-intensive tasks in heterogeneous network scenarios, the cloud offloading mechanism is established for mobile end-users with the quality of service constraints, and online learning algorithms are studied to obtain near-optimal offloading decisions. This project tries to design a new extensible architecture and a series of service optimization frameworks for the existing mobile service architecture, improving the scalability and service quality of the large-scale mobile cloud service network.