基于Decision Transformer模型的5G基站节能控制策略优化新范式研究

夏鹏程; 朱银涛; 于霄; 何怡; 倪艺洋

doi:10.11959/j.issn.2096-3750.2025.00481

您当前的位置：

首页 >

文章列表页 >

基于Decision Transformer模型的5G基站节能控制策略优化新范式研究

能源物联网 | 更新时间：2025-07-29

- 基于Decision Transformer模型的5G基站节能控制策略优化新范式研究
- Research on the new paradigm for optimizing 5G base station energy-saving control strategies utilizing the Decision Transformer model
- 物联网学报 2025年9卷第2期页码：82-94
- 作者机构：
  
  1.南京理工大学电子工程与光电技术学院，江苏南京 210094
  2.南京理工大学网络空间安全学院，江苏江阴 214443
  3.江苏第二师范学院计算机工程学院，江苏南京 211200
- 作者简介：
  
  [ "夏鹏程（1993‒ ），男，南京理工大学电子工程与光电技术学院博士生，主要研究方向为无线通信系统节能、工业互联网、工业区块链。" ]
  [ "朱银涛（1999‒ ），男，南京理工大学电子工程与光电技术学院硕士生，主要研究方向为无线决策大模型、强化学习、大语言模型、信息与通信系统。" ]
  [ "于霄（2000‒ ），男，南京理工大学电子工程与光电技术学院硕士生，主要研究方向为深度强化学习、无线通信系统、5G基站节能。" ]
  [ "何怡（1998‒ ），男，南京理工大学网络空间安全学院硕士生，主要研究方向为强化学习、5G基站节能决策。" ]
  [ "倪艺洋（1986‒ ），女，博士，江苏第二师范学院计算机工程学院教授，主要研究方向为无线通信、物联网等。" ]
- 基金信息：
  
  国家自然科学基金资助项目(62471204);江苏省基础研究计划（自然科学基金）前沿引领技术基础研究专项(BK20212001);江苏省高等学校基础科学（自然科学）研究重大项目(24KJA510003)
- DOI：10.11959/j.issn.2096-3750.2025.00481
  中图分类号： TN929.53
- 收稿日期：2025-01-04，
  
  修回日期：2025-02-19，
  
  纸质出版日期：2025-06-10
- 稿件说明：
移动端阅览
夏鹏程,朱银涛,于霄等.基于Decision Transformer模型的5G基站节能控制策略优化新范式研究[J].物联网学报,2025,09(02):82-94.

XIA Pengcheng,ZHU Yintao,YU Xiao,et al.Research on the new paradigm for optimizing 5G base station energy-saving control strategies utilizing the Decision Transformer model[J].Chinese Journal on Internet of Things,2025,09(02):82-94.
夏鹏程,朱银涛,于霄等.基于Decision Transformer模型的5G基站节能控制策略优化新范式研究[J].物联网学报,2025,09(02):82-94. DOI： 10.11959/j.issn.2096-3750.2025.00481.

XIA Pengcheng,ZHU Yintao,YU Xiao,et al.Research on the new paradigm for optimizing 5G base station energy-saving control strategies utilizing the Decision Transformer model[J].Chinese Journal on Internet of Things,2025,09(02):82-94. DOI： 10.11959/j.issn.2096-3750.2025.00481.

摘要

随着5G网络的持续深入应用，基站设备的增长势头迅猛，这不仅催生了通信行业对提升5G基站综合能效和实现节能减排的新需求，也对相关厂商提出了更高的要求。虽然传统的强化学习（RL

reinforcement learning）技术有望优化5G基站的节能策略，但此类方法需要大量的环境互动和模型训练时间。此外，面对动态变化的基站运行环境，状态和动作空间的变化也会导致RL难以学习到较好的策略，且传统RL模型的泛化能力也有限。为了解决这些问题，创新性地提出了基于Decision Transformer（DT）模型的5G基站节能控制策略优化新框架，该框架将每个基站对应的状态和动作空间解耦，以适应不同的场景任务，并且基于轨迹先验改进了原始DT模型，通过轨迹数据的先验信息来优化模型的期望回报。仿真实验结果表明，所提方法与其他RL算法相比，可以在保障用户服务质量的前提下大幅降低系统功耗，且能够在不重新训练的情况下适应未知任务，所提方法在5G基站节能决策场景中展现出明显优势和应用潜力。

Abstract

With the ongoing proliferation of 5G networks

there has been a surge in the deployment of base station equipment. This trend has not only given rise to new demands within the telecommunications industry for enhancing the overall energy efficiency of 5G base stations and achieving energy conservation and emission reduction but also imposed higher standards on related manufacturers. While traditional reinforcement learning (RL) techniques hold promise for optimizing energy-saving strategies for 5G base stations

they require extensive environmental interactions and model training time. Moreover

in the face of dynamically changing base station operating environments

the variability in state and action spaces can make it difficult for RL to learn effective strategies

and the generalization ability of traditional RL models is also limited. To address these challenges

a new framework for optimizing 5G base station energy-saving control strategies based on the Decision Transformer (DT) model was innovatively proposed. The framework decouples the state and action spaces corresponding to each base station for different scenario tasks and improves the original DT model based on the trajectory priors to optimize the expected return of the model through the prior information of the trajectory data. Simulation results demonstrate that compared to other RL algorithms

the proposed method can significantly reduce system power consumption while ensuring the quality of service for users

and it can adapt to unknown tasks without retraining

showcasing the distinct advantages and application potential of our approach in the context of 5G base station energy-saving decision-making.

关键词

Keywords

references

PIOVESAN N , LÓPEZ-PÉREZ D , DE DOMENICO A , et al . Machine learning and analytical power consumption models for 5G base stations [J ] . IEEE Communications Magazine , 2022 , 60 ( 10 ): 56 - 62 .

ISRAR A , YANG Q , ISRAR A . Renewable energy provision and energy-efficient operational management for sustainable 5G infrastructures [J ] . IEEE Transactions on Network and Service Management , 2023 , 20 ( 3 ): 2698 - 2710 .

中兴通讯股份有限公司 . PowerPilot 4G/5G网络节能降耗技术白皮书 [R ] . 2020 .

ZTE Corporation . PowerPilot: white paper on energy saving and consumption reducing technologies for 4G/5G networks [R ] . 2020 .

中国移动研究院 . 5G基站节能技术白皮书 [R ] . 2020 .

China Mobile Research Institute . White paper on energy-saving technologies for 5G base stations [R ] . 2020 .

Huawei Technologies Co. , Ltd . 5G power whitepaper [R ] . 2019 .

刘俊勇 , 潘力 , 何迈 . 能源物联网及其关键技术 [J ] . 物联网学报 , 2020 , 4 ( 4 ): 9 - 16 .

LIU J Y , PAN L , HE M . Internet of energy things and its key technologies [J ] . Chinese Journal on Internet of Things , 2020 , 4 ( 4 ): 9 - 16 .

QIU B , MAO S X , XIAO H L , et al . Power-aware user association and renewable energy configuration for the optimized on-grid energy in hybrid-energy heterogeneous cellular networks [J ] . IEEE Systems Journal , 2024 , 18 ( 1 ): 162 - 173 .

RAJAPAKSHA N , MOHAMMADI J , WESEMANN S , et al . Minimizing energy consumption in MU-MIMO via antenna muting by neural networks with asymmetric loss [J ] . IEEE Transactions on Vehicular Technology , 2024 , 73 ( 5 ): 6600 - 6613 .

LEE B M , YANG H . Energy-efficient massive MIMO in massive industrial Internet of things networks [J ] . IEEE Internet of Things Journal , 2022 , 9 ( 5 ): 3657 - 3671 .

ANDERSSON G , VÄSTBERG A , DEVLIC A , et al . Energy efficient heterogeneous network deployment with cell DTX [C ] // Proceedings of the 2016 IEEE International Conference on Communications (ICC) . Piscataway : IEEE Press , 2016 : 1 - 6 .

FRENGER P , HELMERSSON K W . Massive MIMO muting using dual-polarized and array-size invariant beamforming [C ] // Proceedings of the 2021 IEEE 93rd Vehicular Technology Conference (VTC2021-Spring) . Piscataway : IEEE Press , 2021 : 1 - 6 .

PEESAPATI S K G , OLSSON M , MASOUDI M , et al . An analytical energy performance evaluation methodology for 5G base stations [C ] // Proceedings of the 2021 17th International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob) . Piscataway : IEEE Press , 2021 : 169 - 174 .

PERVAIZ H , ONIRETI O , MOHAMED A , et al . Energy-efficient and load-proportional eNodeB for 5G user-centric networks: a multilevel sleep strategy mechanism [J ] . IEEE Vehicular Technology Magazine , 2018 , 13 ( 4 ): 51 - 59 .

AGGARWAL A , SELVAMUTHU D . Energy optimization of a base station using Q-learning algorithm [C ] // Proceedings of the 2023 17th International Conference on Telecommunications (ConTEL) . Piscataway : IEEE Press , 2023 : 1 - 5 .

MASOUDI M , SOROUSH E , ZANDER J , et al . Digital twin assisted risk-aware sleep mode management using deep Q-networks [J ] . IEEE Transactions on Vehicular Technology , 2023 , 72 ( 1 ): 1224 - 1239 .

WU D , XU Y T , JENKIN M , et al . Energy saving in cellular wireless networks via transfer deep reinforcement learning [C ] // Proceedings of the GLOBECOM 2023-2023 IEEE Global Communications Conference . Piscataway : IEEE Press , 2023 : 7019 - 7024 .

WU Q , CHEN X , ZHOU Z , et al . Deep reinforcement learning with spatio-temporal traffic forecasting for data-driven base station sleep control [J ] . IEEE/ACM Transactions on Networking , 2021 , 29 ( 2 ): 935 - 948 .

WANG X Y , LYU B C , GUO C , et al . A base station sleeping strategy in heterogeneous cellular networks based on user traffic prediction [J ] . IEEE Transactions on Green Communications and Networking , 2024 , 8 ( 1 ): 134 - 149 .

LOTFOLAHI A , FERNG H W . A multi-agent proximal policy optimized joint mechanism in mmWave HetNets with CoMP toward energy efficiency maximization [J ] . IEEE Transactions on Green Communications and Networking , 2024 , 8 ( 1 ): 265 - 278 .

ZHANG J , LI J , WANG Z , et al . Decision transformer for wireless communications: a new paradigm of resource management [J ] . IEEE Wireless Communications , 2025 , 32 ( 2 ): 180 - 186 .

CHEN L L , LU K , RAJESWARAN A , et al . Decision transformer: reinforcement learning via sequence modeling [C ] // Advances in Neural Information Processing Systems (NeurIPS) . Cambridge : MIT Press , 2021 : 15084 - 15097 .

LEE K H , NACHUM O , YANG M J , et al . Multigame decision transformers [C ] // Advances in Neural Information Processing Systems (NeurIPS) . Cambridge : MIT Press , 2022 : 27921 - 27936 .

XIA P C , ZANG M F , ZHAO J , et al . Decision transformer-based efficient data offloading in LEO-IoT [J ] . Entropy , 2024 , 26 ( 10 ): 846 .

XU M D , SHEN Y K , ZHANG S , et al . Prompting decision transformer for few-shot policy generalization [EB ] . arXiv preprint , 2022 , arXiv: 2206.13499 .

NAWAR M , SAAD A , ELHALAWANY B M , et al . Decision transformer-powered traffic control system [C ] // Proceedings of the 2023 11th International Japan-Africa Conference on Electronics, Communications, and Computations (JAC-ECC) . Piscataway : IEEE Press , 2023 : 138 - 141 .

曾德泽 , 李跃鹏 , 赵宇阳 , 等 . 基于强化学习的高能效基站动态调度方法 [J ] . 计算机科学 , 2021 , 48 ( 11 ): 363 - 371 .

ZENG D Z , LI Y P , ZHAO Y Y , et al . Reinforcement learning based dynamic basestation orchestration for high energy efficiency [J ] . Computer Science , 2021 , 48 ( 11 ): 363 - 371 .

XU F L , LI Y , WANG H D , et al . Understanding mobile traffic patterns of large scale cellular towers in urban environment [J ] . IEEE/ACM Transactions on Networking , 2017 , 25 ( 2 ): 1147 - 1161 .

3GPP. Radio access network: study on channel model for frequencies from 0.5 to 100 GHz v15.0.0：TR 38.901 [S ] . 2018 .

温醒醒 , 贾亦真 , 黄康 . 基于异构蜂窝业务量预测的基站休眠策略 [J ] . 重庆邮电大学学报(自然科学版) , 2021 , 33 ( 6 ): 921 - 929 .

WEN X X , JIA Y Z , HUANG K . Base station sleeping strategy based on heterogeneous cellular traffic prediction [J ] . Journal of Chongqing University of Posts and Telecommunications (Natural Science Edition) , 2021 , 33 ( 6 ): 921 - 929 .

TOMBAZ S , FRENGER P , OLSSON M , et al . Energy performance of 5G-NX radio access at country level [C ] // Proceedings of the 2016 IEEE 12th International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob) . Piscataway : IEEE Press , 2016 : 1 - 6 .

WANG Y F , YANG C , WEN Y , et al . Critic-guided decision transformer for offline reinforcement learning [J ] . Proceedings of the AAAI Conference on Artificial Intelligence , 2024 , 38 ( 14 ): 15706 - 15714 .

CHEN Z Y , HU J , MIN G Y . Learning-based resource allocation in cloud data center using advantage actor-critic [C ] // Proceedings of the ICC 2019 - 2019 IEEE International Conference on Communications (ICC) . Piscataway : IEEE Press , 2019 : 1 - 6 .

MENG W J , ZHENG Q , SHI Y , et al . An off-policy trust region policy optimization method with monotonic improvement guarantee for deep reinforcement learning [J ] . IEEE Transactions on Neural Networks and Learning Systems , 2022 , 33 ( 5 ): 2223 - 2235 .

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于强化学习的能量采集传感器传输功率控制方法

无人机集群协同主动搜索的强化学习策略研究

基于强化学习的多基站协作接收时隙Aloha网络信道接入机制

5G多站定位精度分析与提升方法

5G网络时敏业务的时延特性分析