面向差异化服务的雾无线接入网络协同传输理论与方法

The service industry that relies on information technology has become an important focal point for the optimization and upgrade of China's industrial structure. As the bridge among all kinds of services, however, there are still exists some significant challenges in the traditional fog radio access networks (F-RANs), including the uncertainty of cooperation transmission performance causes, the difficulty of matching the service requirements and the network resource, as well as the complexity of the multi-dimension resource management methods. These issues cause that the traditional F-RAN could not satisfy the requirements of differentiation services and have become a serious obstacle to the advance of China's information technology strategy. To address the above mentioned problems, the fundamental theory should be broken through, and based on our previous research, this project will conduct research on 1) To reap all the benefits of F-RANs, the dynamical cooperation theory for differentiation services in F-RANs will be modeled, and the theoretical limits of performance and the performance scaling laws will be exploited; 2) In order to satisfy the different performance requirements of the services, the cooperative information sensing methods based on data-driven will be proposed, which can flexibly match the service requirements and network resource; 3) Due to the high computational complexity and poor versatility of the traditional resource allocation schemes, the multi-dimension resource allocation methods based on deep reinforcement learning will be proposed, which can allocate the multi-dimension resources effectively and further improve the network performance. The research of this project is expected to break through the bottlenecks of cooperation transmission theory in F-RANs, and have a positive impact on the research and application in China's information service industry. In addition, the achievements of this research will provide the theoretical foundation for the evolution of the future radio access network architecture.

依靠信息技术的服务产业已成为推进我国产业结构优化升级的重要着力点。然而作为各类服务的核心纽带，传统F-RAN中尚存在协同传输性能成因不明确，服务需求与网络资源难以适配，多维资源优化方法复杂度高的技术挑战，不能有效支撑未来差异化服务，严重制约着我国信息化战略推进。针对以上问题，本课题拟从基础理论上突破创新，在已有基础上研究：①构建面向差异化服务的F-RAN动态协同传输理论模型，分析理论性能极限和关键成因，从而深入挖掘网络性能潜力；②为满足差异化服务不同性能需求，研究基于数据驱动的协同感知方法，实现差异化服务需求与所需资源的灵活适配；③针对F-RAN多维资源紧耦合求解复杂的难点，提出基于机器学习的智能资源优化方法，实现多维资源的高效调配和性能整体提升。本项目研究有望突破目前F-RAN协同传输理论瓶颈，对信息化服务产业的研究与应用产生积极作用，同时可为下一代无线接入网络的演进发展提供理论参考。

本项目针对传统雾无线接入网络(F-RAN)中存在的协同传输性能成因不明确，服务需求与网络资源难以适配，多维资源优化方法复杂度高的技术挑战展开了研究：①构建了面向差异化服务的F-RAN动态协同传输理论模型，分析了通信-感知、通信-计算间性能成因；②研究了基于数据驱动的协同感知方法，实现差异化服务需求与所需资源的灵活适配，满足了差异化服务不同性能需求；③针对F-RAN多维资源紧耦合求解复杂的难点，提出了基于机器学习的智能资源优化方法，实现了多维资源的高效调配和性能整体提升。本项目研究共发表论文13篇，其中一作/通讯作者SCI论文7篇；申请发明专利4项，其中已授权1项；出版学术专著1项；获IEEE/CIC ICCC国际会议最佳论文奖、北京市科学技术奖一等奖。本项目以我国未来信息化发展战略为导向，取得了一批具有原创性的达到国际先进水平的科研成果，为F-RAN标准应用提供了理论支撑和技术指导。