| 摘要: |
| 超密集网络(Ultra-dense Network,UDN)中集成移动边缘计算(Mobile Edge Computing,MEC)，是5G中为用户提供计算资源的可靠方式，在多种因素影响下进行MEC任务卸载决策一直都是一个研究热点。目前已存在大量任务卸载相关的方案，但是这些方案中很少将重心放在用户在不同条件下的能耗需求差异上，无法有效提升用户体验质量(Quality of Experience,QoE)。在动态MEC系统中提出了一个考虑用户能耗需求的多用户任务卸载问题，通过最大化满意度的方式提升用户QoE，并将现有的深度强化学习算法进行了改进，使其更加适合求解所提优化问题。仿真结果表明，所提算法较现有算法在算法收敛性以及稳定性上具有一定提升。 |
| 关键词: 超密集网络(UDN) 移动边缘计算（MEC） 卸载方案 深度强化学习 |
| DOI：10.20079/j.issn.1001-893x.220113001 |
|
| 基金项目:国家自然科学基金资助项目(62071077) |
|
| Dynamic task offloading based on MEC in ultra-dense networks |
| XIAN Yongju,LIU Chuang,HAN Ruiyin,CHEN Wanqiong |
| (School of Communication and Information Engineering,Chongqing University of Posts and Telecommunications,Chongqing 400065,China) |
| Abstract: |
| The integration of mobile edge computing(MEC) in ultra-dense network(UDN) is a reliable way to provide users with computing resources in 5G.The decision of MEC task offloading under the influence of many factors has always been a research hotspot.At present,there are many schemes related to task offloading,but few of them focus on the difference of users energy consumption demand under different conditions,which can not effectively improve the quality of experience(QoE).In the dynamic MEC system,a multi-user task offloading problem considering users energy consumption demand is proposed,which improves users QoE by maximizing satisfaction,and improves the existing deep reinforcement learning algorithm to make it more suitable for solving the proposed optimization problem.Simulation results show that the proposed algorithm has a certain improvement in convergence and stability compared with the existing algorithms. |
| Key words: ultra-dense network(UDN) mobile edge computing(MEC) offloading scheme deep reinforcement learning |
