基于强化学习的无线带宽与功率联合资源分配方法研究  

Reinforcement Learning Based Joint Allocation Method for Wireless Bandwidth Scheduling and Power Allocation

作　　者：孙万飞 吴立臣 张晓康 王胡成 徐晖[1,2,3] SUN Wanfei;WU Lihe;ZHANG Xiaokang;ANG uheng;XU Huia;无(CICT Mobile Communication Technology Co.,Ltd.,Beijing 100083,China;State Key Lab of Wireless Mobile Communication,China Academy of Telecommunications Technology,Beijing 100191,China;Datang Mobile Communications Equipment Co.,Ltd.,Beijing 100083,China;Beihang University,Beijing 100191,China)

机构地区：[1]中信科移动通信技术股份有限公司,北京100083 [2]无线移动通信国家重点实验室,电信科学技术研究院,北京100191 [3]大唐移动通信设备有限公司,北京100083 [4]北京航空航天大学,北京100191

出　　处：《移动通信》2024年第6期81-85,共5页Mobile Communications

基　　金：国家重点研发计划“6G智简使能关键技术研究”(2022YFB2902103)。

摘　　要：无线资源调度是移动通信网络研究的核心问题之一,高效可靠的无线资源调度对于服务质量保障和系统效率起到关键作用。面向6G智能内生网络,提出一种基于双延迟深度确定性策略梯度强化学习的PRB分配和功率分配统一调度的方法,相比于传统资源调度和基于强化学习的单用户资源调度,该算法在调度周期内对用户需求和网络资源统一考量,相比于带宽轮询分配和功率平均分配,单PRB速率提升10%以上,有效提升了无线资源的利用效率和系统吞吐量。Wireless resource scheduling is a fundamental issue in mobile communication network research,Efficient and reliable wireless resource scheduling plays a pivotal role in service quality assurance and system efficiency.Towards the development of 6G intelligent endogenous networks,we propose a unified scheduling method for PRB allocation and power allocation based on Twin Delayed Deep Deterministic Policy Gradient(TD3)reinforcement learning in this paper.In contrast to traditional resource scheduling and single-user resource scheduling based on reinforcement learning,this paper's algorithm considers the user's demand and network resources in a unifed manner throughout the scheduling cycle.Furthermore,it is more efficient and reliable than bandwidth polling allocation and power averaging.In comparison to bandwidth polling allocation and power average allocation,the system throughput is enhanced by over 10%,thereby effectively improving the utilisation efficiency of wireless resources and system throughput.

关 键 词：TD3 无线资源 PRB分配 功率分配 联合资源调度 

分 类 号：TN929.5[电子电信—通信与信息系统]