移动边缘计算中资源供给不确定下的最优投资和定价问题  

Optimal investment and pricing under uncertain resource supply in mobile edge computing

作　　者：林德铭 林姿琼 郑艺峰[1,2] 杨敬民 张文杰[1,2] Lin Deming;Lin Ziqiong;Zheng Yifeng;Yang Jingmin;Zhang Wenjie(School of Computer Sciences,Minnan Normal University,Zhangzhou,363000,China;Fujian Provincial University Key Laboratory of Data Science and Intelligent Application,Minnan Normal University,Zhangzhou,363000,China;Department of Electronic Engineering,Taipei University of Technology,Taipei,106344,China)

机构地区：[1]闽南师范大学计算机学院,漳州363000 [2]数据科学与智能应用福建省高校重点实验室,闽南师范大学,漳州363000 [3]台北科技大学电子工程系,台北106344

出　　处：《南京大学学报（自然科学版）》2024年第3期416-428,共13页Journal of Nanjing University（Natural Science）

基　　金：国家自然科学基金(62141602);福建省自然科学基金(2021J011002,2021J011004,2020J01813);教育部人文社科青年项目(20YJC870002)。

摘　　要：为了缓解能源性能的瓶颈,满足更多用户的资源需求,将移动边缘计算(Mobile Edge Computing,MEC)与能量收集(Energy Harvesting,EH)技术相结合,其中,EH技术可以通过捕获环境中的绿色能量来促进设备的可持续计算,提出一个具有随机能量收集的EH-MEC系统,以研究资源供给不确定性的投资和定价问题.用户可以租赁MEC系统的固有资源,也可以使用EH装置收集的资源.由于无线环境的时变性,EH收集的能源具有不确定性,所以,在MEC系统中找到MEC服务器和用户之间的平衡是值得研究的问题.针对这一问题,提出一种序贯决策方法,将用户的投资和MEC服务器资源定价的交互行为建模为一个四阶段的Stackelberg博弈,利用逆向归纳法得到用户与MEC服务器在利润最大化下的纳什均衡解,证明MEC服务器的最优能量收集时间、最优租赁资源和定价决策遵循良好的阈值结构.实验结果表明,绿色资源获取可以显著提高MEC服务器和用户的预期收益.To alleviate the bottleneck of energy performance and meet the resource demand of more users,Mobile Edge Computing(MEC)combined with Energy Harvesting(EH)technology has been considered,in which EH technology facilitates sustainable computing in devices by capturing green energy from environment.Thus,a stochastic EH⁃MEC system is proposed to study the investment and pricing problem with uncertain resource supply.Users can either rent the inherent resources of MEC system or use the resources harvested by EH.Due to the time⁃varying nature of wireless environment,the energy collected by EH has the feature of uncertainty.Therefore,how to find a balance between MEC servers and users in MEC systems is a problem worth studying.To address this problem,a sequential decision⁃making method is proposed,and the interaction between users'investment and MEC server's resource pricing is formulated as a four⁃stage Stackelberg game.Then,backward induction is used to obtain Nash equilibrium for the users and MEC server under profit maximization.The MEC server optimal energy collection time,optimal leasing resources and pricing decisions has been demonstrated to follow a good threshold structure.Experimental results show that green resource acquisition can significantly improve the expected revenue of MEC servers and users.

关 键 词：移动边缘计算 STACKELBERG博弈 能量收集 资源分配 投资与定价 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]