Dynamic Admission Control and Resource Allocation for Mobile Edge Computing Enabled Small Cell Network
Dynamic Admission Control and Resource Allocation for Mobile Edge Computing Enabled Small Cell Network
复制标题
支持移动边缘计算的小型蜂窝网络的动态准入控制和资源分配
DOI：
10.1109/tvt.2021.3133696
复制
发表时间：
2022-02-01
影响因子：
6.8
通讯作者：
Shen, Xuemin
中科院分区：
文献类型：
--
作者：
Huang, Jiwei;Lv, Bofeng;Shen, Xuemin
移动边缘计算(MEC)是近年来兴起的满足终端设备日益增长的资源需求的一种很有前途的模式。同时，采用MEC的小蜂窝网络(SCN)被认为是下一代无线网络的关键组成部分，可以处理指数级增长的数据流量，提高网络覆盖范围。然而，随着越来越多的终端设备需要将计算负载转移到边缘服务器，网络将严重拥塞，从而导致多个设备之间的性能下降和不平衡。本文提出了MEC支持的SCN中的联合接纳控制和计算资源分配问题，并将其描述为一个随机优化问题。目标是最大化系统效用，在限制队列的同时结合吞吐量和公平性。我们将原问题分解为三个相互独立的子问题，可以在不需要系统统计信息的情况下以分布式方式求解。设计了一种接纳控制和计算资源分配(ACRA)算法来获得子问题的最优解。理论分析证明，Accra算法可以获得接近最优的系统效用，并在效用和队列长度之间达到任意折衷。通过实验验证了分析结果，并对阿克拉算法的性能进行了评估。
Mobile edge computing (MEC) has recently risen as a promising paradigm to meet the increasing resource requirements of the terminal devices. Meanwhile, small cell network (SCN) with MEC has been emerging to handle the exponentially increasing data traffic and improve the network coverage, and is recognized as one key component of the next generation wireless networks. However, with the growing number of terminal devices requiring computation offloading to the edge servers, the network would be heavily congested and thus the performance would be degraded and unbalanced among multiple devices. In this paper, we propose the joint admission control and computation resource allocation in the MEC enabled SCN, and formulate it as a stochastic optimization problem. The goal is to maximize the system utility combining the throughput and fairness while bounding the queue. We decouple the original problem into three independent subproblems, which can be solved in a distributed manner without requiring the system statistical information. An admission control and computation resource allocation (ACCRA) algorithm is designed to obtain the optimal solutions of the subproblems. Theoretical analysis proves that the ACCRA algorithm can achieve the close-to-optimal system utility and reach the arbitrary tradeoff between the utility and the queue length. Experiments are conducted to validate the derived analytical results and evaluate the performance of the ACCRA algorithm.