深度强化学习与移动通信资源管理:算法、进展与展望  

Deep Reinforcement Learning and Mobile Communication Resource Management:Algorithms,Progress,and Prospects

作　　者：孙恩昌 袁永仪 吴兵 屈晗星 张延华 SUN Enchang;YUAN Yongyi;WU Bing;QU Hanxing;ZHANG Yanhua(Faculty of Information Technology,Beijing University of Technology,Beijing 100124,China;Beijing Laboratory of Advanced Information Networks,Beijing 100124,China;Beijing-Dublin International College,Beijing University of Technology,Beijing 100124,China)

机构地区：[1]北京工业大学信息学部,北京100124 [2]先进信息网络北京实验室,北京100124 [3]北京工业大学北京-都柏林国际学院,北京100124

出　　处：《北京工业大学学报》2023年第1期71-88,共18页Journal of Beijing University of Technology

基　　金：国家自然科学基金资助项目(61671029);中国国家留学基金高等学校骨干教师研修项目(2018-10038);北京市博士后工作经费资助项目(ZZ2019-73)。

摘　　要：深度强化学习(deep reinforcement learning,DRL)将深度学习从高维数据提取低维特征的能力与强化学习的决策能力相结合,是移动通信资源管理与优化的高效算法之一.在引入DRL相关算法概念与原理的基础上,重点对DRL在网络切片、云计算、雾计算、移动边缘计算等通信技术与场景中的资源管理与优化效果进行综述与分析,结合DRL在移动通信资源管理的算法原理与研究进展,论述了DRL面临的问题与挑战,并提出相应解决思路.最后,展望了DRL在移动通信资源管理领域的发展趋势和主要研究方向.As one of the highly-efficient algorithms for resource management and optimization in mobile communications,deep reinforcement learning(DRL)integrates the ability of deep learning to extract low dimensional features from high dimensional data with the decision-making ability of reinforcement learning.First,the concepts and principles of DRL algorithms were introduced.Then,the resource management and optimization effect of DRL in different scenarios were summarized and analyzed.The technologies and scenarios included network slicing,cloud computing,fog computing,and mobile edge computing.Furthermore,based on the key research progress of DRL in mobile communication resource management,the open issues and challenges of DRL were discussed,and possible solutions were proposed.Finally,development trends and key research directions in the field of mobile communication resoure management were prospected.

关 键 词：深度强化学习(deep reinforcement learning DRL) 通信资源管理 网络切片 云计算 雾计算 移动边缘计算 

分 类 号：TN929.5[电子电信—通信与信息系统]