当前位置: X-MOL 学术arXiv.cs.NI › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
CL-ADMM: A Cooperative Learning Based Optimization Framework for Resource Management in MEC
arXiv - CS - Networking and Internet Architecture Pub Date : 2020-03-28 , DOI: arxiv-2003.12846
Xiaoxiong Zhong, Xinghan Wang, Li Li, Yuanyuan Yang, Yang Qin, Tingting Yang, Bin Zhang, Weizhe Zhang

We consider the problem of intelligent and efficient resource management framework in mobile edge computing (MEC), which can reduce delay and energy consumption, featuring distributed optimization and efficient congestion avoidance mechanism. In this paper, we present a Cooperative Learning framework for resource management in MEC from an Alternating Direction Method of Multipliers (ADMM) perspective, called CL-ADMM framework. First, in order to caching task efficiently in a group, a novel task popularity estimating scheme is proposed, which is based on semi-Markov process model, then a greedy task cooperative caching mechanism has been established, which can effectively reduce delay and energy consumption. Secondly, for addressing group congestion, a dynamic task migration scheme based on cooperative improved Q-learning is proposed, which can effectively reduce delay and alleviate congestion. Thirdly, for minimizing delay and energy consumption for resources allocation in a group, we formulate it as an optimization problem with a large number of variables, and then exploit a novel ADMM based scheme to address this problem, which can reduce the complexity of problem with a new set of auxiliary variables, these sub-problems are all convex problems, and can be solved by using a primal-dual approach, guaranteeing its convergences. Then we prove that the convergence by using Lyapunov theory. Numerical results demonstrate the effectiveness of the CL-ADMM and it can effectively reduce delay and energy consumption for MEC.

中文翻译:

CL-ADMM:基于合作学习的 MEC 资源管理优化框架

我们考虑了移动边缘计算(MEC)中智能高效的资源管理框架问题,该框架可以减少延迟和能耗,具有分布式优化和高效拥塞避免机制。在本文中,我们从乘法器交替方向法 (ADMM) 的角度提出了一个用于 MEC 资源管理的合作学习框架,称为 CL-ADMM 框架。首先,为了在组内高效缓存任务,提出了一种新的基于半马尔可夫过程模型的任务流行度估计方案,然后建立了贪婪任务协同缓存机制,可以有效降低延迟和能耗. 其次,针对群体拥塞,提出了一种基于协同改进Q-learning的动态任务迁移方案,可以有效减少延迟,缓解拥塞。第三,为了最小化组内资源分配的延迟和能量消耗,我们将其表述为一个具有大量变量的优化问题,然后利用一种新的基于 ADMM 的方案来解决这个问题,这可以降低问题的复杂度一组新的辅助变量,这些子问题都是凸问题,可以用原对偶方法求解,保证收敛。然后我们用李雅普诺夫理论证明收敛性。数值结果证明了CL-ADMM的有效性,它可以有效地减少MEC的延迟和能耗。我们将其表述为具有大量变量的优化问题,然后利用一种新的基于 ADMM 的方案来解决该问题,该方案可以通过一组新的辅助变量来降低问题的复杂度,这些子问题都是凸问题,并且可以通过使用原始对偶方法解决,保证其收敛。然后我们用李雅普诺夫理论证明收敛性。数值结果证明了CL-ADMM的有效性,它可以有效地减少MEC的延迟和能耗。我们将其表述为具有大量变量的优化问题,然后利用一种新的基于 ADMM 的方案来解决该问题,该方案可以通过一组新的辅助变量来降低问题的复杂度,这些子问题都是凸问题,并且可以通过使用原始对偶方法解决,保证其收敛。然后我们用李雅普诺夫理论证明收敛性。数值结果证明了CL-ADMM的有效性,它可以有效地减少MEC的延迟和能耗。保证其收敛。然后我们用李雅普诺夫理论证明了收敛性。数值结果证明了CL-ADMM的有效性,它可以有效地减少MEC的延迟和能耗。保证其收敛。然后我们用李雅普诺夫理论证明收敛性。数值结果证明了CL-ADMM的有效性,它可以有效地减少MEC的延迟和能耗。
更新日期:2020-06-09
down
wechat
bug