5G wireless artificial intelligence dataset construction analysis

doi:10.12267/j.issn.2096-5931.2025.03.011

Abstract

Abstract:

With the rapid development of Artificial Intelligence (AI) technology, the introduction of AI-based design into 5G air interface design has become an important direction for the future evolution of 5G. The establishment of 5G wireless AI dataset is a key step for introducing AI technology into 5G air interface design. This paper provides a systematic analysis of the establishment of 5G wireless AI dataset. Firstly, it elaborates the basic theories and methods involved in establishing 5G wireless AI dataset. Then, based on the latest progress in 3GPP research, it conducts an analysis and demonstration on the methods for establishing datasets for several typical use cases, including AI-based Channel State Information (CSI) feedback, beam management and positioning enhancement. Finally, it proposes the prospects for the continuous evolution of 5G wireless AI datasets.

Key words: 5G, AI, dataset

CLC Number:

TP309.2
TP18

LIU Xiaofeng. 5G wireless artificial intelligence dataset construction analysis[J]. Information and Communications Technology and Policy, 2025, 51(3): 83-88.

Add to citation manager EndNote|Ris|BibTeX

URL:

http://ictp.caict.ac.cn/EN/10.12267/j.issn.2096-5931.2025.03.011

http://ictp.caict.ac.cn/EN/Y2025/V51/I3/83

参数FDD,OFDM	数值双工方式及波形
场景	UMi、Uma等
频率范围	FR1,2 GHz或4 GHz
基站间距/m	200
信道建模方式	基于TR 38.901^[6]
基站天线配置	32端口:(8,8,2,1,1,2,8),(dH,dV)=(0.5,0.8)λ
终端天线配置	4接收天线:(1,2,2,1,1,1,2),(dH,dV)=(0.5,0.5)λ
基站发送功率	41 dBm(10 MHz)、44 dBm(20 MHz)、47 dBm(40 MHz)
基站天线高度/m	25
终端接收底噪/dB	9
基础参数	时隙配置:每时隙14 OFDM符号载波间隔:15 kHz
带宽/MHz	10
多天线层数	最大天线层数可配置,如8、12
终端分布	80%室内(3 km/h),20%室外(30 km/h)
信道估计方式	真实信道估计

参数FDD,OFDM	数值双工方式及波形
场景	UMi、Uma等
频率范围	FR1,2 GHz或4 GHz
基站间距/m	200
信道建模方式	基于TR 38.901^[6]
基站天线配置	32端口:(8,8,2,1,1,2,8),(dH,dV)=(0.5,0.8)λ
终端天线配置	4接收天线:(1,2,2,1,1,1,2),(dH,dV)=(0.5,0.5)λ
基站发送功率	41 dBm(10 MHz)、44 dBm(20 MHz)、47 dBm(40 MHz)
基站天线高度/m	25
终端接收底噪/dB	9
基础参数	时隙配置:每时隙14 OFDM符号载波间隔:15 kHz
带宽/MHz	10
多天线层数	最大天线层数可配置,如8、12
终端分布	80%室内(3 km/h),20%室外(30 km/h)
信道估计方式	真实信道估计

主要指标	主要配置	非AI方案	AI方案
余弦相似度	单层低负载传输	0.7 513	0.8 301
	单层高负载传输	0.8 223	0.9 108
	双层低负载传输	0.7 128	0.7 428
	双层高负载传输	0.7 385	0.7 754

主要指标	主要配置	非AI方案	AI方案
余弦相似度	单层低负载传输	0.7 513	0.8 301
	单层高负载传输	0.8 223	0.9 108
	双层低负载传输	0.7 128	0.7 428
	双层高负载传输	0.7 385	0.7 754

参数	数值
场景	密集城区Uma场景,7 站, 每站3扇区
载波频率/GHz	30
子载波间隔/kHz	120
系统带宽/MHz	80
基站发送功率/dBm	40
基站间距/m	200
基站天线配置	[Mg,Ng,M,N,P] = [1,1,4,8,2],[dV,dH] = [0.5,0.5] λ
终端天线配置	[Mg,Ng,M,N,P] = [1,2,1,4,2],[dV,dH] = [0.5,0.5] λ
基站发送天线模式	32发送天线:水平角度= [-78.75,-56.25,-33.75,-11.25,11.25,33.75,56.25,78.75];垂直角度= [22.5,67.5,112.5,157.5]
终端接收天线模式	每个面板4接收天线:水平角度= [-67.5,-22.5,22.5,67.5]
终端移动速度/km/h	3、30等
其他因素	终端翻转,空间一致性

5G wireless artificial intelligence dataset construction analysis

RichHTML

PDF

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 6

References 6

Related Articles 15

Recommended Articles

Metrics

Comments

基于AI波束管理方案	最优波束预测准确率
256波束选16波束随机选取	36.6%
256波束选16波束固定均匀选取	56.99%
256波束选32波束随机选取	55.59%
256波束选32波束固定均匀选取	74.81%

[1]	HE H, JIN S, WEN C K, et al. Model-driven deep learning for physical layer communications[J]. IEEE Wireless Commun, 2019, 26(5):77-83.
[2]	WEN C K, SHIH W T, JIN S. Deep learning for massive MIMO CSI feedback[J]. IEEE Wireless Commun, 2018, 7(5):748-751.
[3]	XIAO H, WANG Z Q, TIAN W Q, et al. AI enlightens wireless communication:analyses,solutions and opportunities on CSI feedback[J]. 中国通信:英文版, 2021, 18(11):104-116.
[4]	王志勤, 刘晓峰, 沈嘉, 等. 5G+AI融合全景图[M]. 北京: 人民邮电出版社, 2022.
[5]	3GPP. Study on artificial intelligence (AI)/machine learning (ML) for NR air interface: RP-213599[S/OL]. 2022[2023-09-20]. https://www.3gpp.org/ftp/tsgran/TSGRAN/TSGR94e/Docs/RP-213599.zip.
[6]	3GPP. Study on channel model for frequencies from 0.5 to 100 GHz (Release 17): TS 38.901, V17.0.0[S], 2022.

[1]	FAN Xuemei, MAN Jianzhi, ZHANG Hui, LONG Shanli, SUN Guangyu. Research on motor imagery brain-computer interface rehabilitation training method incorporating attention assessment and feedback [J]. Information and Communications Technology and Policy, 2025, 51(3): 25-35.
[2]	HUANG Cen, LIU Yuying, HUANG Shaolong, LI Mengtian, PANG Shu, LIN Weibin, LI Fang. Exploration on the future industry development of brain science and brain-computer engineering: a case study of Shenzhen [J]. Information and Communications Technology and Policy, 2025, 51(3): 44-52.
[3]	ZHOU Jie, CHENG Liwei. Global brain computer interface technology and industry development trends [J]. Information and Communications Technology and Policy, 2025, 51(3): 53-58.
[4]	MU Xiangxu, JIA Mingyan, WANG Tiancheng, WU Meixi. High-quality computing power trends in the era of artificial intelligence [J]. Information and Communications Technology and Policy, 2025, 51(2): 18-23.
[5]	LI Qin, HAN Liuyan, LI Han, ZHANG Dechao. Research on new technology of SPN for electric power applications [J]. Information and Communications Technology and Policy, 2025, 51(2): 70-75.
[6]	SHANG Xingyu, LIU Xiao’ou, YANG Mingchuan. Study on artificial intelligence native network [J]. Information and Communications Technology and Policy, 2025, 51(2): 9-17.
[7]	WANG Shida, YANG Bingyi. India advances Industry 4.0 with artificial intelligence technology [J]. Information and Communications Technology and Policy, 2025, 51(1): 71-75.
[8]	MA Yan. AIGC technology empowers the value and path of digital twin city construction [J]. Information and Communications Technology and Policy, 2024, 50(9): 43-50.
[9]	YANG Yueqiang, CHEN Jie, LIU Qian. Research on the standardization status and key technologies of broadband customer network security capability in the F5G era [J]. Information and Communications Technology and Policy, 2024, 50(9): 85-91.
[10]	DONG Yaocong, ZHANG Qian, LI Baoqiang, LI Yi, DONG Yue. Research on security technology and application of industrial internet based on generative artificial intelligence [J]. Information and Communications Technology and Policy, 2024, 50(8): 32-37.
[11]	YUCHI Xuebiao, YUAN Weiguo, LI Hongtao. Research on internet domain name abuse information disposal architecture [J]. Information and Communications Technology and Policy, 2024, 50(8): 38-45.
[12]	ZHOU Linzhi, CHEN Chen, ZHENG Haoran, SHI Yi, XING Jiaming, LIN Fengxu. Design and implementation of email access control gateway for abnormal behavior [J]. Information and Communications Technology and Policy, 2024, 50(8): 46-54.
[13]	WENG Chenxun, LI Mingyang, YIN Hualei, CHEN Zengbing. Quantum Byzantine agreement in blockchain: protocol, application and outlook [J]. Information and Communications Technology and Policy, 2024, 50(7): 59-68.
[14]	ZHANG Qian, HUANG Huang, MIAO Zicong. Research on development trend of key technologies and industry of intelligent computing [J]. Information and Communications Technology and Policy, 2024, 50(6): 10-16.
[15]	DU Yanyan, YAO Minqi, CHEN Wenqi. Comprehensive application practice of all factor connected vehicles industry based on 5G/5G-A [J]. Information and Communications Technology and Policy, 2024, 50(6): 62-69.

模型类型	模型大小/ M	计算复杂度/ MFLOPs	90%用户位置预测准确度/m
CNN	2.4	4.8	0.65
CNN+ResNet	2.76	1 694	0.046

模型类型	模型大小/ M	计算复杂度/ MFLOPs	90%用户位置预测准确度/m
CNN	2.4	4.8	0.65
CNN+ResNet	2.76	1 694	0.046