边缘增强H-CRAN中能耗感知的资源分配机制  

Resource allocation mechanism with energy consumption awareness in the edge enhanced H-CRAN

作　　者：吕翊[1,2,3] 王焰斌 张鸿 王汝言[1,2,3] 张普宁 LV Yi;WANG Yanbin;ZHANG Hong;WANG Ruyan;ZHANG Puning(School of Communication and Information Engineering,Chongqing University of Posts and Telecommunications,Chongqing 400065,China;Chongqing Key Laboratory of Optical Communication and Networks,Chongqing 400065,China;Chongqing Key Laboratory of Ubiquitous Sensing and Networking,Chongqing 400065,China)

机构地区：[1]重庆邮电大学通信与信息工程学院,重庆400065 [2]重庆高校市级光通信与网络重点实验室,重庆400065 [3]泛在感知与互联重庆市重点实验室,重庆400065

出　　处：《西安电子科技大学学报》2022年第2期67-78,107,共13页Journal of Xidian University

基　　金：国家自然科学基金(61771082,61871062,61901071,U20A20157);重庆市高校创新研究群体(CXQT20017);重庆市教委科学技术研究项目(KJQN202000626,KJQN201800615);重庆市自然科学基金面上项目(cstc2019jcyj-msxmX0303);重庆市自然科学基金重点项目(cstc2020jcyj-zdxmX0024)。

摘　　要：针对目前移动边缘计算增强异构云无线接入网中的设备能耗高、资源利用率低以及用户服务质量恶化等不足之处,从频谱资源与计算资源的角度出发,提出了一种基于能耗感知的通信与计算资源分配机制。首先将网络吞吐量作为收入,能耗作为成本支出,建立起了服务提供商视角的利润模型框架。为了避免由于资源分配不均导致的边缘服务器资源浪费或过载,首先通过分析用户到来的各项服务请求,并利用稀疏矩阵算法合理分配频谱资源使网络吞吐量得到提升,而针对计算资源则设计了一种启发式算法,以确定用户关联和用户计算资源需求量,使每个边缘服务器都能得到充分利用;并基于资源利用的结果并结合考虑光纤前传链路容量约束,可以动态部署移动边缘计算服务器于宏基站或者远端射频头处以降低设备开销。针对不同参数指标与一天不同时刻服务请求情况的仿真结果表明,所提机制能有效地提高网络吞吐量、减少网络能耗并降低光纤前传链路的阻塞概率,相较于其他算法在各项指标上均有显著提升。Focusing on the shortcomings of low resource utilization rate,high equipment energy consumption and deterioration of user service quality in the mobile edge computing enhanced heterogeneous cloud radio access network at present,an energy consumption aware communication and computing resource allocation mechanism is proposed from the perspective of spectrum resources and computing resources.First,taking the network throughput as the revenue and energy consumption as the cost expenditure,a profit model framework from the perspective of service providers is established.In order to avoid the waste or overload of edge server resources caused by uneven resource allocation,the network throughput is first improved by analyzing various service requests coming from users and reasonably allocating spectrum resources by using the sparse matrix algorithm.For computing resources,a heuristic algorithm is designed to determine user association and user computing resource demand,so that each edge server can be fully utilized.Based on the results of resource utilization and considering the capacity constraints of the optical fiber forward link,the mobile edge computing server can be dynamically deployed at the macro base station or remote radio heads to reduce the equipment overhead.Simulation results for different parameter indexes and service requests at different times of a day show that the proposed mechanism can effectively increase network throughput,reduce network energy consumption and decrease the blocking probability of the optical fiber forward link,so that this mechanism is apparently superior to other algorithms.

关 键 词：异构云无线接入网 节能 移动边缘计算 资源分配 

分 类 号：TN915[电子电信—通信与信息系统]