AI使能的5G节能技术

doi:10.11959/j.issn.1000-0801.2021104

摘要/Abstract

摘要：

随着 5G 商用的全面开展，5G 无线站点数目急剧增加，5G 核心网需分层部署在区域/省/地市数据中心，以及数据中心规模化发展，导致能耗问题日益凸显。基于全网能耗主要占比，调研5G接入网络、核心网络和数据中心的能源效率评估方法。介绍了AI使能的基站节能技术及试点应用方案、AI应用于 5G 核心网的节能方式、AI 使能的数据中心节能技术和试点应用方案，探讨了节能技术的挑战和未来的研究方向。对整体通信系统节能技术的总结和展望，有助于提高对能源效率和绿色网络发展的认识。

关键词: 5G, AI, 能源效率, 接入网络, 核心网络, 数据中心

Abstract:

With the development of 5G commercialization, the number of 5G wireless base station sites has increased sharply, 5G core networks need to be deployed in regions/provinces/cities data centers, as well as the data centers were developing on a larger scale, the problem of energy consumption was becoming increasingly prominent.Based on the main proportion of energy consumption in the whole network, the energy efficiency evaluation methods of 5G access network, core network and data center was investigated.The AI-enabled energy-saving technology of base stations and pilot application scheme, 5G core network energy-saving methods by using AI, as well as the AI-enabled energy-saving technology of data centers and pilot application scheme were introduced, and the challenges and future research directions of energy-saving technology were finally discussed.The summary and prospect on the energy-saving technology of the overall communication system, will help to improve the understanding of energy efficiency and green network development.

Key words: 5G, AI, energy efficiency, access network, core network, data center

中图分类号:

TN929.5

徐丹, 曾宇, 孟维业, 李力卡. AI使能的5G节能技术[J]. 电信科学, 2021, 37(5): 32-41.

Dan XU, Yu ZENG, Weiye MENG, Lika LI. AI-enabled 5G energy-saving technology[J]. Telecommunications Science, 2021, 37(5): 32-41.

图/表 1

参考文献 50

[1]	华为技术有限公司. 5G 时代运营商数据和存储架构白皮书[R]. 2020.
	Huawei Technologies Co.,Ltd. White paper on operater data and storage architecture in 5G era[R]. 2020.
[2]	GEORGAKOPOULOS A , MARGARIS A , TSAGKARIS K ,et al. Resource sharing in 5G contexts: achieving sustainability with energy and resource efficiency[J]. IEEE Vehicular Technology Magazine, 2016,11(1): 40-49.
[3]	GAO H Y , SU Y M , ZHANG S B ,et al. Antenna selection and power allocation design for 5G massive MIMO uplink networks[J]. China Communications, 2019,16(4): 1-15.
[4]	赛迪顾问. 5G产业发展白皮书（2020）[R]. 2020.
	CCID Consulting. 5G industry development white paper (2020)[R]. 2020.
[5]	岑祺 . 5G 基站市电建设及改造方案[J]. 信息通信, 2019,32(12): 210-213.
	CEN Q . 5G base station municipal power construction and transformation plan[J]. Information ＆ Communications, 2019,32(12): 210-213.
[6]	WU Q Q , LI G Y , CHEN W ,et al. An overview of sustainable green 5G networks[J]. IEEE Wireless Communications, 2017,24(4): 72-80.
[7]	OH E , KRISHNAMACHARI B , LIU X ,et al. Toward dynamic energy-efficient operation of cellular network infrastructure[J]. IEEE Communications Magazine, 2011,49(6): 56-61.
[8]	MEMON M L , MAHESHWARI M K , SAXENA N ,et al. Artificial intelligence-based discontinuous reception for energy saving in 5G networks[J]. Electronics, 2019,8(7): 778.
[9]	LOPEZ-PEREZ D , DE DOMENICO A , PIOVESAN N ,et al. A survey on 5G energy efficiency: massive MIMO,lean carrier design,sleep modes,and machine learning[J]. arXiv preprint arXiv:2101.11246, 2021.
[10]	LORINCZ J , CAPONE A , WU J S . Greener,energy-efficient and sustainable networks: state-of-the-art and new trends[J]. Sensors (Basel,Switzerland), 2019,19(22): E4864.
[11]	LI Y N R , CHEN M Z , XU J ,et al. Power saving techniques for 5G and beyond[J]. IEEE Access, 2020(8): 108675-108690.
[12]	3GPP. Study on new aspects of energy efficiency (EE) for 5G:TR 28.813 V0.5.0[S]. 2021.
[13]	ITU-T. Total network infrastructure energy efficiency metrics:L.1332[S]. 2018.
[14]	CHAO L , QOUNEH A , TAO L . iSwitch: coordinating and optimizing renewable energy powered server clusters[J]. ACM Sigarch Computer Architecture News, 2012,40(3): 512-523.
[15]	DEBAILLIE B , DESSET C , LOUAGIE F . A flexible and future-proof power model for cellular base stations[C]// Proceedings of 2015 IEEE 81st Vehicular Technology Conference (VTC Spring). Piscataway: IEEE Press, 2015: 1-7.
[16]	DONEVSKI I , VALLERO G , MARSAN M A . Neural networks for cellular base station switching[C]// Proceedings of IEEE INFOCOM 2019 - IEEE Conference on Computer Communications Workshops (INFOCOM WKSHPS). Piscataway: IEEE Press, 2019: 738-743.
[17]	张志荣, 许晓航, 朱雪田 ,等. 基于AI的5G基站节能技术研究[J]. 电子技术应用, 2019,45(10): 1-4.
	ZHANG Z R , XU X H , ZHU X T ,et al. Research on energy saving technology of 5G base station based on AI[J]. Applica-tion of Electronic Technique, 2019,45(10): 1-4.
[18]	LI R P , ZHAO Z F , ZHOU X ,et al. Intelligent 5G: when cellular networks meet artificial intelligence[J]. IEEE Wireless Communications, 2017,24(5): 175-183.
[19]	GAO Y , CHEN J J , LIU Z ,et al. Machine learning based energy saving scheme in wireless access networks[C]// Proceedings of 2020 International Wireless Communications and Mobile Computing (IWCMC). Piscataway:IEEE Press, 2020: 1573-1578.
[20]	SOLIMAN S S , SONG B . Fifth generation (5G)cellular and the network for tomorrow: cognitive and cooperative approach for energy savings[J]. Journal of Network and Computer Applications, 2017,85: 84-93.
[21]	SDN/NFV/AI标准与产业推进委员会. 智能化基站节能研究[R]. 2020.
	SDN/NFV/AI Standards and Industry Promotion Committee. Research on energy saving of intelligent base station[R]. 2020.
[22]	张青 . 基于AI的5G基站节能场景智能识别[C]// 5G网络创新研讨会（2019）论文集. 移动通信, 2019(Z2): 123-127.
	ZHANG Q , . Intelligent identification of energy saving scenario of 5G base station based on AI[C]// Symposium on 5G Net-work Innovation (2019). Mobile Communication, 2019(Z2): 123-127.
[23]	任嘉鹏 . 基于机器学习的流量预测及基站休眠方法研究[D]. 长春:吉林大学, 2020.
	REN J P . Research on machine learning based traffic prediction and base station sleep mode[D]. Changchun:Jilin University, 2020.
[24]	李继蕊, 李小勇, 高云全 ,等. 5G 网络下移动云计算节能措施研究[J]. 计算机学报, 2017,40(7): 1491-1516.
	LI J R , LI X Y , GAO Y Q ,et al. Energy saving research on mo-bile cloud computing in 5G[J]. Chinese Journal of Computers, 2017,40(7): 1491-1516.
[25]	易芝玲, 孙奇, 吴杰 ,等. 人工智能在5G无线网络中的标准与应用进展[J]. 信息通信技术与政策, 2020(9): 23-30.
	CHIH-LIN I , SUN QI , WU JIE ,et al. AI in 5G networks—usage scenario and standardization progress[J]. Information and Communications Technology and Policy, 2020(9): 23-30.
[26]	ZHANG X , YOU J L . A gated dilated causal convolution based encoder-decoder for network traffic forecasting[J]. IEEE Access, 2020(8): 6087-6097.
[27]	HOSSAIN M S , MUHAMMAD G . A deep-tree-model-based radio resource distribution for 5G networks[J]. IEEE Wireless Communications, 2020,27(1): 62-67.
[28]	WANG J , TANG J , XU Z Y ,et al. Spatiotemporal modeling and prediction in cellular networks: a big data enabled deep learning approach[C]// Proceedings of IEEE INFOCOM 2017 - IEEE Conference on Computer Communications. Piscataway: IEEE Press, 2017: 1-9.
[29]	ZHANG C , ZHANG H , YUAN D ,et al. Citywide cellular traffic prediction based on densely connected convolutional neural networks[J]. IEEE Communications Letters, 2018,22(8): 1656-1659.
[30]	ZHANG C Y , PATRAS P . Long-term mobile traffic forecasting using deep spatio-temporal neural networks[C]// Proceedings of the Eighteenth ACM International Symposium on Mobile Ad Hoc Networking and Computing. New York:ACM Press, 2018: 231-240.
[31]	WANG X , ZHOU Z M , XIAO F ,et al. Spatio-temporal analysis and prediction of cellular traffic in metropolis[J]. IEEE Transactions on Mobile Computing, 2019,18(9): 2190-2202.
[32]	AL-QUZWEENI A N , LAWEY A Q , ELGORASHI T E H ,et al. Optimized energy aware 5G network function virtualization[J]. IEEE Access, 2019(7): 44939-44958.
[33]	AL-QUZWEENI A , EL-GORASHI T E H , NONDE L ,et al. Energy efficient network function virtualization in 5G networks[C]// Proceedings of 2015 17th International Conference on Transparent Optical Networks (ICTON). Piscataway: IEEE Press, 2015: 1-4.
[34]	KIM H G , LEE D Y , JEONG S Y ,et al. Machine learning-based method for prediction of virtual network function resource demands[C]// Proceedings of 2019 IEEE Conference on Network Softwarization (NetSoft). Piscataway:IEEE Press, 2019: 405-413.
[35]	XU J , WANG J Y , QI Q ,et al. IARA: an intelligent application-aware VNF for network resource allocation with deep learning[C]// Proceedings of 2018 15th Annual IEEE International Conference on Sensing,Communication,and Networking (SECON). Piscataway: IEEE Press, 2018: 1-3.
[36]	REN Y , PHUNG-DUC T , CHEN J C ,et al. Dynamic auto scaling algorithm (DASA) for 5G mobile networks[C]// Proceedings of 2016 IEEE Global Communications Conference (GLOBECOM). Piscataway: IEEE Press, 2016: 1-6.
[37]	徐丹, 王海宁 . AI助力5G发展研究[J]. 中国电信建设, 2019,31(3): 46-54.
	XU D , WANG H N . AI helps 5G development research[J]. China Telecommunications Construction, 2019,31(3): 46-54.
[38]	吴亚奇 . 数据中心机房节能方法研究[D]. 苏州:苏州科技大学, 2019.
	WU Y Q . Research on energy saving method of data center computer room[D]. Suzhou:Suzhou University of Science and Technology, 2019.
[39]	邹燕 . IDC 机房温度调控设计研究[D]. 上海:上海工程技术大学, 2016.
	ZOU Y . The reserch on the control of IDC room temperature[D]. Shanghai:Shanghai University of Engineering Science, 2016.
[40]	屈毅, 宁铎, 赖展翅 ,等. 温室温度控制系统的神经网络PID 控制[J]. 农业工程学报, 2011,27(2): 307-311.
	QU Y , NING D , LAI Z C ,et al. Neural networks based on PID control for greenhouse temperature[J]. Transactions of the Chi-nese Society of Agricultural Engineering, 2011,27(2): 307-311.
[41]	张志鹏, 龚慧钦, 陈庆 . 基于人工智能的机房空调控制方法、装置及系统:CN110285532A[P].2019-09-27.
	ZHANG Z P , GONG H Q , CHEN Q . Machine room air condi-tioner control method,device and system based on artificial in-telligence:CN110285532A[P].2019-09-27.
[42]	徐慧姣 . 数据中心机房的节能减排技术及发展[J]. 通讯世界, 2019,26(3): 251-259.
	XU H J . Energy saving and emission reduction technology and development of data center machine room[J]. Telecom World, 2019,26(3): 251-259.
[43]	周鹏飞, 马亮 . 一种基于人工智能优化机房能耗效率的方法及装置:CN109890176A[P].2019-06-14.
	ZHOU P F , MA L . Method and device for optimizing machine room energy consumption efficiency on the basis of artificial intelligence:CN109890176A[P].2019-06-14.
[44]	杨震, 赵静洲, 林依挺 ,等. 数据中心 PUE 能效优化的机器学习方法[J]. 系统工程理论与实践, 2020: 1-12.
	YANG Z , ZHAO J Z , LIN Y T ,et al. Machine learning ap-proach for energy efficiency optimization of PUE in data cen-ters[J]. Systems Engineering-Theory ＆ Practice, 2020: 1-12.
[45]	娄洁良 . 数据中心能效案例及节能运行[J]. 电信科学, 2019,35(2): 95-104.
	LOU J L . Data center energy efficiency case and energy saving operation[J]. Telecommunications Science, 2019,35(2): 95-104.
[46]	BUZZI S , I C L , KLEIN T E ,et al. A survey of energy-efficient techniques for 5G networks and challenges ahead[J]. IEEE Journal on Selected Areas in Communications, 2016,34(4): 697-709.
[47]	CHEN T , YANG Y , ZHANG H G ,et al. Network energy saving technologies for green wireless access networks[J]. IEEE Wireless Communications, 2011,18(5): 30-38.
[48]	KAUR J , KHAN M A , IFTIKHAR M ,et al. Machine learning techniques for 5G and beyond[J]. IEEE Access, 2021(9): 23472-23488.
[49]	KAIROUZ E B P , MCMAHAN H B . Advances and open problems in federated learning[J]. Foundations and Trends? in Machine Learning, 2021,14(1).Doi:10.1561/2200000083.
[50]	PRECHELT L . Early stopping-but when?[M]// Lecture Notes in Computer Science. Berlin,Heidelberg: Springer Berlin Heidelberg, 2012: 53-67.

模型名称	模型特点描述	参考文献
改进的LSTM模型	考虑不同基站的流量之间的空间依赖性，改进LSTM的结构	参考文献[26]
基于树的深度模型	基于树的模型使用了卷积层，可以有效捕获流量的空间依赖信息	参考文献[27]
混合的时空预测深度学习模型	时间依赖性被LSTM捕获，而空间依赖性被自动编码器封装，用于建模来自相邻基站的历史流量信息，从而将流量的空间依赖性纳入预测	参考文献[28]
流量预测架构	通过利用密集连接的中枢神经网络来捕获时空依赖关系	参考文献[29]
基于卷积神经网络的预测模型	通过建模时间和长距离的空间依赖关系，该模型遵循了一个编码器对应一个解码器的范式，组合了LSTM和CNN元素	参考文献[30]
基于图神经网络（graph neural network， GNN）	利用图表示对网络流量时空依赖进行建模。将总数据流量分解为基站内流量和基站间流量，这分别对应于基站覆盖范围内的用户服务流量和在不同建站覆盖区域之间移动的用户服务流量	参考文献[31]