双端可虚拟化的移动接入网架构理论及算法研究

In response to the real-world demands from mobile operators, mobile end users and mobile service providers, this project proposes a new mobile access network architecture that supports virtualization at two ends: networks and mobile device (also called user equipment or UE). The corresponding design theory and resource allocation algorithms will be investigated. .Firstly, this architecture emphasizes on the equal importance of UEs. It aims to expanding the promising C-RAN architecture by introducing UE virtualization at the application layer and UE cooperative communications in a device-to-device (D2D) manner at the lower layers. This solution will be able to improve the energy-spectrum efficiency on one hand and to prolong UE battery’s lifetime on the other hand. A QoS-constrained app offloading strategy and its corresponding mathematical expression will be investigated. The constraints include maximum delay for user applications and the wireless resource availability, etc..Secondly, a holistic research into UE cooperative D2D communications and networking will be carried out for both single cell and multiple cells. Join resource allocation will be conducted for both power and radio spectrum in addition to multi-hop relay selection. For this purpose, an information-theoretic model will be established first, which will be followed by optimization algorithms and their detailed analysis in terms of solvability and computational complexity. Corresponding algorithms for reducing the complexity will also be investigated..Thirdly, a set of cross-layer resource allocation algorithms will be proposed that joint UE application offloading with UE cooperative communication, aiming to further improve energy-spectrum efficiency. The algorithms will be optimized by using various modern optimization techniques. Not confined to only simulations the project also aims to setting up an experimental test-bed and to implementing a prototype system to validate its novel design.

针对移动运营商、移动用户及服务提供商的实际需求，研究一种双端虚拟化的移动接入网的新架构、新机理和跨层的无线资源管理算法。首先，该架构突出用户设备（UE）的重要性，从UE虚拟化和UE协作直通通信两个方面拓展C-RAN架构，给出系统的理论模型，提出QoS约束的移动应用卸载准则及其数学描述，以进一步提高无线网络的能谱效率，并在一定程度上解决手机电池使用时间短的问题。其次，给出完整的网络通信层UE协作中继理论和算法，包括单小区和多小区。在明确给出D2D网络的频谱和能量资源的数学定义的基础上建立D2D网络能谱资源联合优化的数学模型，并对数学问题的可解性和求解复杂度进行分析，设计低复杂度问题求解算法。最后，实现跨层的高能谱效率的联合资源管理。将应用层的UE应用卸载和下层的多中继多小区UE协作通信联合设计，建立联合优化数学模型并研究高效算法。本项目不限于仿真，同时搭建实验环境，实现原型系统。

本项目针对移动运营商频谱资源受限、移动用户手机电池使用时间短、及服务提供商对某些实时服务低延迟的实际需求，提出了一种双端虚拟化的移动接入网的新架构、以及在此基础上的新机理和跨层的无线资源管理算法。..首先，该架构突出用户设备（UE）的重要性，从UE虚拟化和UE协作通信两个方面拓展目前的以C-RAN为代表的移动网络架构，给出了系统的整体建模，提出QoS约束的移动应用卸载准则及其数学描述和决策算法，进一步提高了无线网络的能谱效率，并在一定程度上解决手机电池使用时间短的问题。..其次，给出完整的网络通信层UE协作中继理论和算法，包括单小区和多小区。在明确给出D2D网络的频谱和能量资源的数学定义的基础上建立了D2D网络能谱资源联合优化的数学模型，并对数学问题的可解性和求解复杂度进行分析，设计低复杂度问题求解算法。将应用层的UE应用卸载和下层的动态通信资源分配联合设计，建立联合优化数学模型并设计出了高效算法。..最后，实现了基于5G开源平台OAI的移动网络的搭建，实现了UE在移动边缘云端的虚拟化，并设计和实现了2个演示场景：面向大计算量的UE任务卸载和面向大数据传输的视频转码。..这些理论结果和实际平台给下一代通信网络的设计提供了宝贵的信息和指导，尤其在移动网络虚拟化方面具有较大的指导意义，拓展了网络虚拟化的范围到用户终端，并在移动核心网中提供支持和统一管理。