大规模移动边缘计算网络:空间建模及计算吞吐量优化  

Large-Scale Mobile Edge Computing Network:Spatial Modelingand Computation Throughput Optimization

作　　者：韩凯峰 胡昌军 刘铁志 HAN Kaifeng;HU Changjun;LIU Tiezhi(China Academy of Information and Communications Technology,Beijing 100191,China)

机构地区：[1]中国信息通信研究院,中国北京100191

出　　处：《中兴通讯技术》2020年第4期6-12,共7页ZTE Technology Journal

摘　　要：提出并定义了大规模移动边缘计算(MEC)网络中的空间计算吞吐量这一性能指标,通过运用随机几何、凸优化等理论,对这一性能指标进行了分析和最优化设计。利用随机几何理论为大规模MEC网络建立空间模型,该模型涵盖边缘云及用户的空间随机分布、无线接入、计算任务卸载、边缘端并行计算等重要的网络特征。基于网络模型,首次对MEC网络空间计算吞吐量进行定义和分析,并通过优化设计MEC服务范围半径(r0)以及用户计算卸载比例(ρ)这两个指标,来实现MEC网络空间吞吐量的最大化。所提供的严谨的理论分析、富有物理内涵的优化结果将为部署大规模MEC网络提供了极为重要的设计指南。The attempt on defining,analyzing,and optimizing the spatial computation throughput in a large-scale wireless mobile edge computing(MEC)network is made.The analysis involves the interplay of theories of stochastic geometry and convex optimization.Specifically,the large-scale MEC network features of wireless access and edge-computing are first modeled,such as random nodes distribution,computation tasks offloading,parallel computing at edges,by using stochastic geometry.Based on the proposed model,the spatial computation throughput of the MEC network is defined,studied and maximized in terms of the radius of MEC service range(denoted by r0)as well as offloading ratio of active mobile users(denoted byρ)under the constraints of latency and energy.The optimal solutions of r0 andρcan be easily calculated via solving simple equations,and their closed-form results could be obtained in the extreme case.The tractable analysis and insightful results give useful design guidelines for MEC network planning and provisioning in a large-scale space.

关 键 词：移动边缘计算 移动计算卸载 无线网络建模 随机几何 凸优化 

分 类 号：TN929.5[电子电信—通信与信息系统]