Resource Modeling and Scheduling for Mobile Edge Computing: A Service Provider’s Perspective
Resource Modeling and Scheduling for Mobile Edge Computing: A Service Provider’s Perspective
复制标题
DOI：
10.1109/access.2018.2851392
复制
发表时间：
2018-06
期刊：
影响因子：
3.9
通讯作者：
Shuaishuai Guo;Dalei Wu;Haixia Zhang;D. Yuan
中科院分区：
文献类型：
--
作者：
Shuaishuai Guo;Dalei Wu;Haixia Zhang;D. Yuan
研究了面向移动的边缘计算（MEC）服务的基站（BS）资源建模与管理问题。在所提出的建模中，BS被视为由多个多类型服务器组成的排队网络。联合考虑具有不同优先级的上行传输用户、下行传输用户和MEC用户。假设它们的服务请求动态地到达并且也动态地被服务。有了这样一个通用的资源模型，用户之间的交互就可以基于网络理论进行分析。推导了不同优先级的业务类型的平均时延。基于推导的结果，两个资源管理优化问题，制定和解决的角度来看，服务提供商。MEC服务带来的收入首先通过进行用户准入控制来最大化，同时用给定量的通信和计算资源来提供所有准入用户的服务质量（QoS）。然后，通过满足所有用户的QoS，使资源部署的资本支出最小化。它被公式化为一个整数规划问题。该算法可以帮助服务提供商确定最优的通信和计算资源量，以最小的总资本支出来保证所有用户的QoS。计算机模拟验证了所有的分析和比较，与BS服务的单一优先级的多类型的用户。通过比较，发现服务提供商可以通过区分用户优先级来获得更多的收入或节省更少的资本支出。
This paper investigates resource modeling and management for a base station (BS) providing mobile edge computing (MEC) service. In the proposed modeling, BS is recognized as a queueing network consisting of multiple multi-type servers. The uplink transmission users, downlink transmission users, and MEC users with different priority levels are jointly considered. It is assumed that their service-requests arrive dynamically and are also served dynamically. With such a general resource modeling, the interaction among these users can be analyzed based on the queueing network theory. The average delay of each service-type with different priority levels is derived. Based on the derived results, two resource management optimization problems are formulated and solved from the perspective of a service provider. The revenue brought by MEC services is first maximized by doing user admission control while provisioning the quality-of-service (QoS) of all admitted users with the given amount of communication and computation resources. Then, the capital expenditure of resource deployment is minimized by satisfying the QoS of all users. It is formulated as an integer programming problem. An algorithm is developed to solve it, which can help service providers to determine the optimal amount of communication and computation resources to be placed in a BS to guarantee QoS for all users at a minimal total capital expenditure. Computer simulations are done to validate all analysis and comparisons are made with BS serving multi-type users of single priority level. Through comparison, an insight is gained that service providers can obtain more revenue or spare less capital expenditure by differentiating the user priority levels.