基于深度神经网络的铁路通信系统频谱感知方法

袁张付; 宋宗莹; 王兴中; 周一鸣

doi:10.11959/j.issn.1000-0801.2025157

您当前的位置：

首页 >

文章列表页 >

基于深度神经网络的铁路通信系统频谱感知方法

研究与开发 | 更新时间：2025-11-16

- 基于深度神经网络的铁路通信系统频谱感知方法
- Spectrum sensing method of railway communication system based on deep neural network
- 电信科学 2025年41卷第9期页码：108-118
- 作者机构：
  
  1.中国神华能源股份有限公司，北京 100011
  2.北京邮电大学电子工程学院，北京 100876
- 作者简介：
  
  [ "袁张付（1979- ），男，现就职于中国神华能源股份有限公司，主要研究方向为移动通信、光传输、核心网通信技术、标准及测试。" ]
  宋宗莹（1981- ），男，博士，中国神华能源股份有限公司正高级工程师，主要研究方向为重载铁路调度、运输、专网、列控技术和智慧化等。
  王兴中（1968- ），男，博士，中国神华能源股份有限公司高级工程师，主要研究方向为列车网络化控制。
  周一鸣（2001- ），男，北京邮电大学电子工程学院硕士生，主要研究方向为信号频谱感知和深度学习。
- 基金信息：
  
  中国神华能源股份有限公司科技项目(SHGF-24-02)
- DOI：10.11959/j.issn.1000-0801.2025157
  中图分类号： TN929.5
- 收稿：2025-02-12，
  
  修回：2025-04-30，
  
  录用：2025-06-03，
  
  纸质出版：2025-09-20
- 稿件说明：
移动端阅览
袁张付,宋宗莹,王兴中等.基于深度神经网络的铁路通信系统频谱感知方法[J].电信科学,2025,41(09):108-118.

YUAN Zhangfu,SONG Zongying,WANG Xingzhong,et al.Spectrum sensing method of railway communication system based on deep neural network[J].Telecommunications Science,2025,41(09):108-118.
袁张付,宋宗莹,王兴中等.基于深度神经网络的铁路通信系统频谱感知方法[J].电信科学,2025,41(09):108-118. DOI： 10.11959/j.issn.1000-0801.2025157.

YUAN Zhangfu,SONG Zongying,WANG Xingzhong,et al.Spectrum sensing method of railway communication system based on deep neural network[J].Telecommunications Science,2025,41(09):108-118. DOI： 10.11959/j.issn.1000-0801.2025157.

摘要

针对当前铁路通信系统400 MHz专用频段中存在的频谱资源紧张问题，提出了一种基于深度神经网络的频谱感知方法。该方法通过联合分析信号的空-频-时联合表征，在保障主用户通信质量的前提下实现高精度频谱空穴检测。具体实现如下：首先，认知用户对接收到的感知样本进行信号处理，提取信号的能量、功率谱和循环谱特征，并拼接为信号特征矩阵，作为神经网络的输入；其次，通过特征嵌入模块对各模态特征进行深层表示；最后，在混合感知层中利用2个多层感知机（multilayer perceptron，MLP）模块分别提取各模态特征及特征间的关联性，并完成频谱判决。实验结果表明，在虚警概率为0.01、信噪比为-10 dB的仿真环境下，该方法的检测概率可达99.8%。与目前基于卷积神经网络和卷积长短期记忆深度神经网络的频谱感知方法相比，该方法的检测概率平均可提升3.23%、2.61%。所提方法为铁路异构通信系统的动态频谱接入提供了高可靠性解决方案。

Abstract

To address the spectrum scarcity issue in the 400 MHz dedicated frequency band of railway communication systems

a deep neural network（DNN）-based spectrum sensing method was proposed. By jointly analyzing the spatial-frequency-temporal joint characterization of signals

high-precision spectrum hole detection was achieved while ensuring the communication quality of primary users. The implementation procedure was conducted as follows: firstly

the received sensing samples were processed by cognitive users

where energy features

power spectral density

and cyclostationary characteristics were extracted and concatenated into a signal feature matrix as the neural network input. Subsequently

each modal feature was deeply represented through a feature embedding module. Finally

two multilayer perceptron (MLP) modules in the hybrid sensing layer were utilized to extract both intra-modal features and inter-modal correlations

followed by spectrum decision-making. Experimental results demonstrate that under simulated conditions with a false alarm probability of 0.01 and signal-to-noise ratio (SNR) of -10 dB

the proposed method achieves a detection probability of 99.8%. Compared with existing spectrum sensing methods based on convolutional neural networks (CNN) and convolutional long short-term memory deep neural networks

the detection probability was improved by 3.23% and 2.61% on average

respectively. This method provides a highly reliable solution for dynamic spectrum access in heterogeneous railway communication systems.

关键词

Keywords

references

ZHONG Z D ， AI B ， ZHU G ， et al . Dedicated mobile communications for high-speed railway [M ] . Heidelberg : Springer ， 2018 .

HE R S , AI B , WANG G P , et al . High-speed railway communications: from GSM-R to LTE-R [J ] . IEEE Vehicular Technology Magazine , 2016 , 11 ( 3 ): 49 - 58 .

YIN Z J , WANG Y M , WU C . A multi-agent collaborative model for Bayesian opportunistic channel accessibility in railway cognitive radio [J ] . International Journal of Performability Engineering , 2017 , 13 ( 4 ): 479 - 489 .

CHAE K , PARK J , KIM Y . Rethinking autocorrelation for deep spectrum sensing in cognitive radio networks [J ] . IEEE Internet of Things Journal , 2023 , 10 ( 1 ): 31 - 41 .

王琳 , 张世龙 , 王树彬 . 基于注意力多尺度特征融合的频谱感知方法 [J ] . 无线电工程 , 2024 , 54 ( 11 ): 2520 - 2526 .

WANG L , ZHANG S L , WANG S B . A spectrum sensing method based on attention multi-scale feature fusion [J ] . Radio Engineering , 2024 , 54 ( 11 ): 2520 - 2526 .

CAI L N , CAO K T , WU Y P , et al . Spectrum sensing based on spectrogram-aware CNN for cognitive radio network [J ] . IEEE Wireless Communications Letters , 2022 , 11 ( 10 ): 2135 - 2139 .

李盼盼 , 吴昊 , 刘佳佳 , 等 . 智能铁路5G安全技术与策略综述 [J ] . 计算机科学 , 2024 , 51 ( 5 ): 1 - 11 .

LI P P , WU H , LIU J J , et al . Overview of security technologies and strategies for intelligent railway 5G [J ] . Computer Science , 2024 , 51 ( 5 ): 1 - 11 .

LIU C , WANG J , LIU X M , et al . Deep CM-CNN for spectrum sensing in cognitive radio [J ] . IEEE Journal on Selected Areas in Communications , 2019 , 37 ( 10 ): 2306 - 2321 .

DIGHAM F F , ALOUINI M S , SIMON M K . On the energy detection of unknown signals over fading channels [C ] // Proceedings of the IEEE International Conference on Communications , 2003 . Piscataway : IEEE Press , 2003: 3575 - 3579 .

ZENG Y , KOH C L , LIANG Y C . Maximum eigenvalue detection: theory and application [C ] // Proceedings of the 2008 IEEE International Conference on Communications . Piscataway : IEEE Press , 2008 : 4160 - 4164 .

SALAHDINE F , EL GHAZI H , KAABOUCH N , et al . Matched filter detection with dynamic threshold for cognitive radio networks [C ] // Proceedings of the 2015 International Conference on Wireless Networks and Mobile Communications (WINCOM) . Piscataway : IEEE Press , 2015 : 1 - 6 .

JANG W M . Blind cyclostationary spectrum sensing in cognitive radios [J ] . IEEE Communications Letters , 2014 , 18 ( 3 ): 393 - 396 .

PAN G L , LI J , LIN F . A cognitive radio spectrum sensing method for an OFDM signal based on deep learning and cycle spectrum [J ] . International Journal of Digital Multimedia Broadcasting , 2020 ( 1 ): 5069021 .

ZHENG S L , CHEN S C , QI P H , et al . Spectrum sensing based on deep learning classification for cognitive radios [J ] . China Communications , 2020 , 17 ( 2 ): 138 - 148 .

XIE J D , FANG J , LIU C , et al . Unsupervised deep spectrum sensing: a variational auto-encoder based approach [J ] . IEEE Transactions on Vehicular Technology , 2020 , 69 ( 5 ): 5307 - 5319 .

LI H Q , BORHANI-DARIAN P , WU P , et al . Deep learning of GNSS signal correlation [C ] // Proceedings of the 33rd International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2020) . Manassas : Institute of Navigation , 2020 : 2836 - 2847 .

PATEL D K , LÓPEZ-BENÍTEZ M , SONI B , et al . Artificial neural network design for improved spectrum sensing in cognitive radio [J ] . Wireless Networks , 2020 , 26 ( 8 ): 6155 - 6174 .

GAO J B , YI X M , ZHONG C J , et al . Deep learning for spectrum sensing [J ] . IEEE Wireless Communications Letters , 2019 , 8 ( 6 ): 1727 - 1730 .

刘思佚 , 徐东辉 , 刘丁胤 , 等 . 基于 CNN-Transformer 融合的频谱感知方法研究 [J ] . 无线电通信术 , 2024 : 1 - 7 .

LIU S Y , XU D H , LIU D Y , et al . Research on spectrum sensing method based on CNN-Transformer fusion [J ] . Radio Communications , 2024 : 1 - 7 .

TOLSTIKHIN I , HOULSBY N , KOLESNIKOV A , et al . MLP-Mixer: an all-MLP architecture for vision [J ] . arXiv preprint , 2021 , arXiv: 2105.01601 .

YU Y X , TENG Y L , WANG B H , et al . M-Net: a lightweight network based on multilayer perceptron for massive MIMO CSI feedback [C ] // Proceedings of the 2023 IEEE Globecom Workshops (GC Wkshps) . Piscataway : IEEE Press , 2023 : 26 - 31 .

GENG Y , HUANG J Y , YANG J X , et al . Spectrum sensing for cognitive radio based on feature extraction and deep learning [J ] . Journal of Physics: Conference Series , 2022 , 2261 ( 1 ): 012016 .

浏览量

423

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Nyquist折叠接收和CBDNet-YOLOv5的非合作信号超宽带感知与载频测量

基于随机共振和非中心F分布的频谱感知算法

基于5GtoB定制网的数据分流方案研究

基于随机共振的双门限协作频谱感知算法

基于随机矩阵特征值比的机会协作频谱感知