伪三维卷积注意力网络的多步信道预测

陶静; 侯萌; 彭薇; 张国彦; 戴佳明; 刘卫明; 王海东; 王臻

doi:10.11999/JEIT251090

伪三维卷积注意力网络的多步信道预测

doi: 10.11999/JEIT251090 cstr: 32379.14.JEIT251090

陶静^{1, 2},
侯萌^1, ,,
彭薇³,
张国彦⁴,
戴佳明⁴,
刘卫明⁴,
王海东⁴,
王臻³

1.
中国电力科学研究院有限公司北京 100192
2.
南京邮电大学通信与信息工程学院南京 210003
3.
华中科技大学电子信息与通信学院武汉 430074
4.
国网内蒙古东部电力有限公司赤峰供电公司赤峰 024099

基金项目: 国家电网有限公司总部科技项目(1400-202455422A-3-5-YS)

详细信息

作者简介:
陶静：女，正高级工程师，研究方向为电算网融合、电力物联网

侯萌：女，工程师，研究方向为电力算力协同

彭薇：女，教授，研究方向为大规模天线智能通信系统

张国彦：男，高级工程师，研究方向为电力系统通信

戴佳明：男，工程师，研究方向为电力系统通信

刘卫明：男，正高级工程师，研究方向为电力系统通信

王海东：男，正高级工程师，研究方向为电力系统通信

王臻：男，硕士研究生，研究方向为电力线通信

通讯作者:
侯萌　houmeng1857@163.com

中图分类号: TN929.5; TP391.4
计量
- 文章访问数: 46
- HTML全文浏览量: 23
- PDF下载量: 3
- 被引次数: 0
出版历程
- 收稿日期: 2025-10-13
- 修回日期: 2025-11-26
- 录用日期: 2025-12-02
- 网络出版日期: 2025-12-09

A Multi-step Channel Prediction Method Based on Pseudo-3D Convolutional Neural Network with Attention Mechanism

1.
China Electric Power Research Institute, Beijing 100192, China
2.
School of Communications and Information Engineering, Nanijing University of Posts and Telecommunications, Nanjing 210003, China
3.
School of Electronic Information and Communication, Huazhong University of Science and Technology, Wuhan 430074, China
4.
Chifeng Power Supply Company, State Grid Inner Mongolia Eastern Electric Power Co., Ltd, Chifeng 024099, China

Funds: The State Grid Corporation of China Headquarters Science and Technology Project (1400-202455422A-3-5-YS)

摘要

摘要: 现有大规模MIMO信道预测多以广义平稳假设为前提，且多采用单步预测。面对非平稳场景，单步结果极易失效，频繁迭代亦显著抬高导频开销。为此，该文构建一套融合伪三维卷积(P3D)与注意力模块的时频联合多步预测框架。该方案以伪三维卷积替代3D卷积实现信道状态信息(CSI)在时域与频域的高效特征提取，并叠加通道与空间的混合注意力(CBAM)，增强网络对全局依赖的表征能力，从而提升预测精度。基于实测信道的实验验证显示，该方法在多步预测任务上具有明显优势。与此同时，结合迁移学习思路，完成了由单天线到多天线场景的平滑扩展。
- 大规模MIMO /
- 多步信道状态信息预测 /
- 伪三维卷积 /
- 混合注意力 /
- 时频联合特征
Abstract: Objective With the rapid growth in connections and data traffic in Fifth Generation (5G) mobile networks, massive Multiple-Input Multiple-Output (MIMO) has become a key technology for improving network performance. The spectral efficiency and energy efficiency of massive MIMO transmission depend on accurate Channel State Information (CSI). However, the non-stationary characteristics of wireless channels, terminal processing delay, and the use of ultra-high-frequency bands intensify CSI aging, which necessitates channel prediction. Most mainstream prediction schemes are designed for generalized stationary channels and rely on single-step prediction. In non-stationary environments, CSI obtained through single-step prediction is likely to become outdated, and frequent single-step prediction greatly increases pilot overhead. To address these challenges, a multi-step channel prediction method based on a Pseudo-Three-Dimensional Convolutional Neural Network (P3D-CNN) and an attention mechanism is proposed. The method learns the joint time–frequency characteristics of CSI, leverages high frequency-domain correlation to mitigate the effect of lower time-domain correlation in multi-step prediction, and improves prediction performance. Methods In this study, the uplink model of a massive MIMO system is constructed (Fig. 1). CSI is obtained through channel estimation, using an Inverse Fast Fourier Transform (IFFT) at the transmitter and a Fast Fourier Transform (FFT) at the receiver. Actual channel measurements provide a CSI dataset with time–frequency dimensions, and autocorrelation analyses are performed in both domains. A multi-step channel prediction network, termed P3D-CNN with the Convolutional Block Attention Module (CBAM) (Fig. 13), is designed. The P3D-CNN structure replaces the traditional Three-Dimensional Convolutional Neural Network (3D-CNN) by decomposing the three-dimensional convolution into a two-dimensional convolution in the frequency domain and a one-dimensional convolution in the time domain, which greatly reduces computational complexity. The CBAM-based hybrid attention mechanism is incorporated to extract global information in the frequency and channel domains, further improving channel prediction accuracy. Results and Discussions Based on the measured CSI dataset, the prediction method using an AutoRegressive (AR) model, the prediction method using Fully Connected Long Short-Term Memory (FC-LSTM), and the prediction method using P3D-CNN-CBAM are compared under different prediction steps. Simulation results show that the average Normalized Mean Square Error (NMSE) of the proposed P3D-CNN-CBAM method is lower than that of the other two methods (Fig. 17). As the prediction step increases from 1 to 10, prediction error rises sharply because the AR model and FC-LSTM rely solely on time-domain correlation. When the prediction step is 10, the average NMSE of these two methods reaches 0.5868 and 0.7648, respectively. The P3D-CNN-CBAM method yields an average NMSE of only 0.3078, maintaining strong prediction performance. The improvement brought by integrating CBAM into the P3D-CNN network is also verified (Fig. 18). Finally, through transfer learning, the proposed method is extended from single-day datasets to multi-day scenarios. Conclusions Based on the measured CSI dataset, a multi-step prediction method addressing CSI aging in massive MIMO systems is proposed. The method applies P3D-CNN with CBAM to improve multi-step prediction accuracy. By replacing full three-dimensional convolution with pseudo-three-dimensional convolution, time–frequency CSI information is effectively extracted, and the CBAM mechanism enhances the learning of global features. Experimental results show that: (1) the proposed method achieves clear performance advantages over AR- and FC-LSTM-based approaches; and (2) through transfer learning, multi-step prediction is extended from single-antenna to multi-antenna scenarios.

HTML全文

图 1 信道测量实验的路线

下载: 全尺寸图片幻灯片

图 2 移动端4天线排列位置

下载: 全尺寸图片幻灯片

图 3 基站端4行8列的双极化天线阵列

下载: 全尺寸图片幻灯片

图 4 实测信道状态信息数据的时域、频域自相关性

下载: 全尺寸图片幻灯片

图 5 P3D-CNN中的空间卷积过程

下载: 全尺寸图片幻灯片

图 6 P3D-CNN中的时域卷积过程

下载: 全尺寸图片幻灯片

图 7 CBAM模块基本结构

下载: 全尺寸图片幻灯片

图 8 通道注意力模块

下载: 全尺寸图片幻灯片

图 9 空间注意力模块

下载: 全尺寸图片幻灯片

图 10 伪3D卷积层与伪3D转置卷积模块结构

下载: 全尺寸图片幻灯片

图 11 P3D-CNN-CBAM网络结构框图

下载: 全尺寸图片幻灯片

图 12 基于FC-LSTM的多步信道预测网络结构框图

下载: 全尺寸图片幻灯片

图 13 基于P3D-CNN-CBAM的多步信道预测方法在单个频点上的5步预测效果

下载: 全尺寸图片幻灯片

图 14 基于P3D-CNN-CBAM的多步信道预测方法在所有频点上的5步预测NMSE结果

下载: 全尺寸图片幻灯片

图 15 各预测步长条件下4种预测方法的平均NMSE统计

下载: 全尺寸图片幻灯片

图 16 各预测步长条件下P3D-CNN与P3D-CNN-CBAM的平均NMSE表现

下载: 全尺寸图片幻灯片

图 17 基站端天线之间的相关性

下载: 全尺寸图片幻灯片

图 18 预测模型训练Loss的变化趋势

下载: 全尺寸图片幻灯片

图 19 通过微调训练获得的基站天线2-5的预测效果

下载: 全尺寸图片幻灯片

参考文献(17)

[1]	Global System for Mobile Communications Assembly (GSMA). The mobile economy China 2024[R]. London: GSMA, 2024.
[2]	REDDY M S, KUMAR T A, and RAO K S. Spectral efficiency analysis of massive MIMO system[C]. 2016 IEEE Conference on Systems, Process and Control (ICSPC), Melaka, Malaysia, 2016: 148–153. doi: 10.1109/SPC.2016.7920720.
[3]	PRASAD K N R S V, HOSSAIN E, and BHARGAVA V K. Energy efficiency in massive MIMO-based 5G networks: Opportunities and challenges[J]. IEEE Wireless Communications, 2017, 24(3): 86–94. doi: 10.1109/MWC.2016.1500374WC.
[4]	HONG W B, BAEK K H, LEE Y, et al. Study and prototyping of practically large-scale mmWave antenna systems for 5G cellular devices[J]. IEEE Communications Magazine, 2014, 52(9): 63–69. doi: 10.1109/MCOM.2014.6894454.
[5]	邱玲. 多用户、多小区MIMO通信技术[M]. 北京: 人民邮电出版社, 2011: 1–290. QIU Ling. Multi-User Multi-Cell MIMO Communication Technology[M]. Beijing: Posts & Telecom Press, 2011: 1–290.
[6]	ZHANG R, ISOLA P, and EFROS A A. Split-brain autoencoders: Unsupervised learning by cross-channel prediction[C]. 2017 IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, USA, 2017: 645–654. doi: 10.1109/CVPR.2017.76.
[7]	郑明明, 彭薇, 刘川, 等. 面向智慧城市数字孪生网络的资源分配方案[J]. 电信科学, 2025, 41(9): 43–54. doi: 10.11959/j.issn.1000-0801.2025210. ZHENG Mingming, PENG Wei, LIU Chuan, et al. Resource allocation scheme for digital twin networks in smart cities[J]. Telecommunications Science, 2025, 41(9): 43–54. doi: 10.11959/j.issn.1000-0801.2025210.
[8]	于创宇, 徐彦彦, 潘少明. 基于MIMO-CSI预测的轻量化信源信道联合编码方法[J]. 电信科学, 2025, 41(6): 29–47. doi: 10.11959/j.issn.1000-0801.2025126. YU Chuangyu, XU Yanyan, and PAN Shaoming. Lightweight joint source-channel coding method based on MIMO-CSI prediction[J]. Telecommunications Science, 2025, 41(6): 29–47. doi: 10.11959/j.issn.1000-0801.2025126.
[9]	LI Xingwang, ZHANG Junyao, HAN Congzheng, et al. Reliability and security of CR-STAR-RIS-NOMA-assisted IoT networks[J]. IEEE Internet of Things Journal, 2024, 11(17): 27969–27980. doi: 10.1109/JIOT.2023.3340371.
[10]	TRUONG K T and HEATH R W. Effects of channel aging in massive MIMO systems[J]. Journal of Communications and Networks, 2013, 15(4): 338–351. doi: 10.1109/JCN.2013.000065.
[11]	李泽慧, 张琳, 山显英. 三维卷积神经网络方法改进及其应用综述[J]. 计算机工程与应用, 2025, 61(3): 48–61. doi: 10.3778/j.issn.1002-8331.2407-0031. LI Zehui, ZHANG Lin, and SHAN Xianying. Review on improvement and application of 3D convolutional neural networks[J]. Computer Engineering and Applications, 2025, 61(3): 48–61. doi: 10.3778/j.issn.1002-8331.2407-0031.
[12]	夏振平, 陈豪, 张宇宁, 等. 基于混合时空卷积的轻量级视频超分辨率重建[J]. 光学精密工程, 2024, 32(16): 2564–2576. doi: 10.37188/OPE.20243216.2564. XIA Zhenping, CHEN Hao, ZHANG Yuning, et al. Lightweight video super-resolution based on hybrid spatio-temporal convolution[J]. Optics and Precision Engineering, 2024, 32(16): 2564–2576. doi: 10.37188/OPE.20243216.2564.
[13]	JIANG Wei, STRUFE M, and SCHOTTEN H D. Long-range MIMO channel prediction using recurrent neural networks[C].2020 IEEE 17th Annual Consumer Communications & Networking Conference (CCNC), Las Vegas, USA, 2020: 1–6. doi: 10.1109/CCNC46108.2020.9045219.
[14]	QIU Zhaofan, YAO Ting, and MEI Tao. Learning spatio-temporal representation with pseudo-3D residual networks[C]. 2017 IEEE International Conference on Computer Vision, Venice, Italy, 2017: 5534–5542. doi: 10.1109/ICCV.2017.590.
[15]	XIE Saining, GIRSHICK R, DOLLÁR P, et al. Aggregated residual transformations for deep neural networks[C]. 2017 IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, USA, 2017: 5987–5995. doi: 10.1109/CVPR.2017.634.
[16]	WOO S, PARK J, LEE J Y, et al. CBAM: Convolutional block attention module[C]. The 15th European Conference on Computer Vision, Munich, Germany, 2018: 3–19. doi: 10.1007/978-3-030-01234-2_1.
[17]	PAN S J and YANG Qiang. A survey on transfer learning[J]. IEEE Transactions on Knowledge and Data Engineering, 2010, 22(10): 1345–1359. doi: 10.1109/TKDE.2009.191.