基于RIS的元素分组面状全连接网络

doi:10.16182/j.issn1004731x.joss.23-0304

系统仿真学报 ›› 2024, Vol. 36 ›› Issue (4): 1017-1027.doi: 10.16182/j.issn1004731x.joss.23-0304

基于RIS的元素分组面状全连接网络

侯顺虎¹(), 方胜良²(), 曾庆尧¹, 王孟涛¹

^1.航天工程大学研究生院，北京 101416
^2.航天工程大学航天信息学院，北京 101416

收稿日期:2023-03-16 修回日期:2023-05-23 出版日期:2024-04-15 发布日期:2024-04-18
通讯作者: 方胜良 E-mail:hshhsc2022@163.com;eeifslyl@163.com
第一作者简介:侯顺虎(2000-)，男，硕士生，研究方向为空间网络电磁安全。E-mail： hshhsc2022@163.com

Element Grouping Faceted Fully Connected Network Based on RIS

Hou Shunhu¹(), Fang Shengliang²(), Zeng Qingyao¹, Wang Mengtao¹

^1.Graduate School, Space Engineering University, Beijing 101416, China
^2.School of Space Information, Space Engineering University, Beijing 101416, China

Received:2023-03-16 Revised:2023-05-23 Online:2024-04-15 Published:2024-04-18
Contact: Fang Shengliang E-mail:hshhsc2022@163.com;eeifslyl@163.com

摘要/Abstract

摘要：

针对神经网络全连接层在训练中参数量多、所占内存多、易产生过拟合问题，从智能超表面(reconfigurable intelligence surface，RIS)结构特征出发，提出了一种基于RIS的元素分组面状全连接神经网络(RIS-based element grouping areal fully connected neural network，RGFCNN)。借鉴RIS的结构特征，在传统全连接神经网络上进行优化。设计了透射面注意力机制用于数据有效特征提取，相比于传统的全连接网络，该网络没有对数据进行一维排列，而是提出了一种运用于神经网络构建的元素分组策略，直接对二维面状数据进行分组全连接处理，各组处理输出进行数据串联。实验结果表明：在公开的具有IQ数据特征的通信信号数据集上，RGFCNN在信噪比大于0 dB时具有更好的识别精度，而训练参数是原来的大约1/6。

关键词: 智能超表面, 全连接神经网络, 元素分组策略, IQ信号, 调制识别

Abstract:

In view of the over-fitting problem that caused by multiple parameters and high memory usage of the full connection layer of neural network in training, a RIS-based element grouping areal fully connected neural network (RGFCNN) is proposed for the first time based on the structural characteristics of reconfigurable intelligence surface (RIS). Based on the structural characteristics of RIS, the network is optimized on traditional FCNN. A novel transmission surface attention mechanism is designed for the effective feature extraction of data. Compared with the traditional FCNNs, the proposed network does not arrange the data in one-dimensional manner. Instead, a element grouping strategy is proposed for the neural network construction, which directly groups the two-dimensional surface data, carries out the fully connected processing on each group, and concatenates the output of each group. The experimental results show that, on the public available communication signal datasets with IQ data features, RGFCNN has better recognition accuracy when SNR is greater than 0 dB, and the training parameters are approximately 1/6 of the original.

Key words: reconfigurable intelligence surface (RIS), fully-connected neural network(FCNN), element grouping strategy, IQ signal, modulation recognition

中图分类号:

TP183

侯顺虎,方胜良,曾庆尧等 . 基于RIS的元素分组面状全连接网络[J]. 系统仿真学报, 2024, 36(4): 1017-1027.

Hou Shunhu,Fang Shengliang,Zeng Qingyao,et al . Element Grouping Faceted Fully Connected Network Based on RIS[J]. Journal of System Simulation, 2024, 36(4): 1017-1027.

图/表 13

图1

图2

图3

图4

表1

图5

表2

图6

图7

图8

图9

表3

图10

参考文献 26

1	赵亚军, 郁光辉, 徐汉青. 6G移动通信网络:愿景、挑战与关键技术[J]. 中国科学(信息科学), 2019, 49(8): 963-987.
	Zhao Yajun, Yu Guanghui, Xu Hanqing. 6G Mobile Communication Networks: Vision, Challenges, and Key Technologies[J]. Scientia Sinica(Informationis), 2019, 49(8): 963-987.
2	Tariq F, Khandaker M R A, Wong K K, et al. A Speculative Study on 6G[J]. IEEE Wireless Communications, 2020, 27(4): 118-125.
3	高子路, 孙韶辉, 李丽. 面向新一代移动通信的智能超表面技术综述[J]. 电信科学, 2022, 38(10): 20-35.
	Gao Zilu, Sun Shaohui, Li Li. Overview of Reconfigurable Intelligent Surface for New-generation Mobile Communication[J]. Telecommunications Science, 2022, 38(10): 20-35.
4	张磊, 崔铁军. 时空编码数字超材料和超表面研究进展[J]. 中国科学基金, 2021, 35(5): 694-700.
	Zhang Lei, Cui Tiejun. Recent Progress of Space-time-coding Digital Metamaterials and Metasurfaces[J]. Bulletin of National Natural Science Foundation of China, 2021, 35(5): 694-700.
5	张磊, 陈晓晴, 郑熠宁, 等. 电磁超表面与信息超表面[J]. 电波科学学报, 2021, 36(6): 817-828.
	Zhang Lei, Chen Xiaoqing, Zheng Yining, et al. Electromagnetic Metasurfaces and Information Metasurfaces[J]. Chinese Journal of Radio Science, 2021, 36(6): 817-828.
6	Yu Nanfang, Genevet P, Kats M A, et al. Light Propagation with Phase Discontinuities: Generalized Laws of Reflection and Refraction[J]. Science, 2011, 334(6054): 333-337.
7	Cui Tiejun, Qi Meiqing, Wan Xiang, et al. Coding Metamaterials, Digital Metamaterials and Programmable Metamaterials[J]. Light: Science & Applications, 2014, 3(10): e218.
8	Mehmet Ali Aygül, Nazzal Mahmoud, Arslan Hüseyin. Deep Learning-based Optimal RIS Interaction Exploiting Previously Sampled Channel Correlations[C]//2021 IEEE Wireless Communications and Networking Conference (WCNC). Piscataway, NJ, USA: IEEE, 2021: 1-6.
9	Xu Meng, Zhang Shun, Ma Jianpeng, et al. Deep Learning-based Time-varying Channel Estimation for RIS Assisted Communication[J]. IEEE Communications Letters, 2022, 26(1): 94-98.
10	朱正. 基于深度学习的太赫兹超表面设计方法研究[D]. 成都: 电子科技大学, 2021.
	Zhu Zheng. Terahertz Metasurface Design Method Based on Deep Learning[D]. Chengdu: University of Electronic Science and Technology of China, 2021.
11	安建成. 面向可重构智能表面的信道估计与被动波束赋形技术研究[D]. 成都: 电子科技大学, 2021.
	An Jiancheng. Channel Estimation and Passive Beamforming for Reconfigurable Intelligent Surface-assisted Communications[D]. Chengdu: University of Electronic Science and Technology of China, 2021.
12	Tang Wankai, Chen Mingzheng, Chen Xiangyu, et al. Wireless Communications with Reconfigurable Intelligent Surface: Path Loss Modeling and Experimental Measurement[J]. IEEE Transactions on Wireless Communications, 2021, 20(1): 421-439.
13	Zhang Lei, Chen Xiaoqing, Liu Shuo, et al. Space-time-coding Digital Metasurfaces[J]. Nature Communications, 2018, 9(1): 4334.
14	Cui Tiejun, Li Lianlin, Liu Shuo, et al. Information Metamaterial Systems[J]. iScience, 2020, 23(8): 101403.
15	Yang Yifei, Zheng Beixiong, Zhang Shuowen, et al. Intelligent Reflecting Surface Meets OFDM: Protocol Design and Rate Maximization[J]. IEEE Transactions on Communications, 2020, 68(7): 4522-4535.
16	Elbir Ahmet M, Mishra K V. A Survey of Deep Learning Architectures for Intelligent Reflecting Surfaces[EB/OL]. (2022-07-21) [2023-01-22]. .
17	Zhang Shunbo, Zhang Shun, Gao Feifei, et al. Deep Learning Optimized Sparse Antenna Activation for Reconfigurable Intelligent Surface Assisted Communication[J]. IEEE Transactions on Communications, 2021, 69(10): 6691-6705.
18	Huang Chongwen, Mo Ronghong, Yuen Chau. Reconfigurable Intelligent Surface Assisted Multiuser MISO Systems Exploiting Deep Reinforcement Learning[J]. IEEE Journal on Selected Areas in Communications, 2020, 38(8): 1839-1850.
19	Feng Keming, Wang Qisheng, Li Xiao, et al. Deep Reinforcement Learning Based Intelligent Reflecting Surface Optimization for MISO Communication Systems[J]. IEEE Wireless Communications Letters, 2020, 9(5): 745-749.
20	Zuo Chao, Qian Jiaming, Feng Shijie, et al. Deep Learning in Optical Metrology: A Review[J]. Light: Science & Applications, 2022, 11(1): 39.
21	Choudhary K, DeCost B, Chen Chi, et al. Recent Advances and Applications of Deep Learning Methods in Materials Science[J]. npj \| Computational Materials, 2022, 8(1): 59.
22	O'Shea T J, West N. Radio Machine Learning Dataset Generation with GNU Radio[C]//Proceedings of the 6th GNU Radio Conference. Tempe, AZ: GNURadio, 2016.
23	Xuan Qi, Li Xiaohui, Chen Zhuangzhi, et al. Deep Transfer Clustering of Radio Signals[EB/OL]. (2021-07-26) [2023-02-19]. .
24	Woo Sanghyun, Park Jongchan, Lee J Y, et al. CBAM: Convolutional Block Attention Module[C]//Computer Vision-ECCV 2018. Cham: Springer International Publishing, 2018: 3-19.
25	Lee Haeyun, Park Jinhyoung, Jae Youn Hwang. Channel Attention Module with Multiscale Grid Average Pooling for Breast Cancer Segmentation in an Ultrasound Image[J]. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, 2020, 67(7): 1344-1353.
26	Du Miao, Yu Qin, Fei Shaomin, et al. Fully Dense Neural Network for the Automatic Modulation Recognition[EB/OL]. (2019-12-07) [2023-03-07]. .

参数	RGFCNN1	RGFCNN2	RGFCNN3	RGFCNN4
输入数据尺寸	(128, 2)	(128, 2)	(128, 2)	(128, 2)
透射层数目	3	3	3	3
元素分组单元尺寸	(4, 2)	(8, 2)	(16, 2)	(32, 2)
元素分组全连接隐藏节点数	4	8	16	32
总的隐藏节点数	128	128	128	128
训练参数量	4 113	4 423	6 423	10 575

参数	FCNN	RGFCNN
输入数据尺寸	(128, 2)	(128, 2)
透射层数目		3
元素分组单元尺寸		(16, 2)
元素分组全连接隐藏节点数		16
总的隐藏节点数	128	128
训练参数量	34 315	6 423

网络名称	分组节点数	总节点数	参数量
FCNN-64	/	64	17 163
RGFCNN-64	8	64	3 607
FCNN-128	/	128	34 315
RGFCNN-128	16	128	6 423
FCNN-256	/	256	68 619
RGFCNN-256	32	256	12 055
FCNN-512	/	512	137 227
RGFCNN-512	64	512	23 319
FCNN-1024	/	1 024	274 443
RGFCNN-1024	128	1 024	45 847

基于RIS的元素分组面状全连接网络

Element Grouping Faceted Fully Connected Network Based on RIS

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 13

参考文献 26

相关文章 1

编辑推荐

Metrics

本文评价