Automatic modulation classification in dynamic noise based on BYOL and real complex domain fusion

Zhao Shaokun; Jia Yong; Zhang Wei; Yao Guangle; Zhang Jian

doi:10.11959/j.issn.1000-0801.DXKX250549

您当前的位置：

首页 >

文章列表页 >

Automatic modulation classification in dynamic noise based on BYOL and real complex domain fusion

Research and Development | 更新时间：2026-06-17

- Automatic modulation classification in dynamic noise based on BYOL and real complex domain fusion
- Telecommunications Science Vol. 42, Issue 5, Pages: 60-73(2026)
- 作者机构：
  
  1.成都理工大学机电工程学院，四川成都 610059
  2.电子科技大学长三角研究院（衢州），浙江衢州 324000
  3.电子科技大学通信与信息工程学院，四川成都 611731
  4.电磁空间安全全国重点实验室，四川成都 610036
- 作者简介：
- 基金信息：
  
  The National Natural Science Foundation of China(U20B2070)
- DOI：10.11959/j.issn.1000-0801.DXKX250549
  CLC： TN92
- Received：12 September 2025，
  
  Revised：2025-11-09，
  
  Accepted：28 November 2025，
  
  Published：20 May 2026
- 稿件说明：
移动端阅览
赵少坤,贾勇,张伟等.基于BYOL与实复域融合的动态噪声自动调制识别[J].电信科学,2026,42(05):60-73.

Zhao Shaokun,Jia Yong,Zhang Wei,et al.Automatic modulation classification in dynamic noise based on BYOL and real complex domain fusion[J].Telecommunications Science,2026,42(05):60-73.
赵少坤,贾勇,张伟等.基于BYOL与实复域融合的动态噪声自动调制识别[J].电信科学,2026,42(05):60-73. DOI： 10.11959/j.issn.1000-0801.DXKX250549.

Zhao Shaokun,Jia Yong,Zhang Wei,et al.Automatic modulation classification in dynamic noise based on BYOL and real complex domain fusion[J].Telecommunications Science,2026,42(05):60-73. DOI： 10.11959/j.issn.1000-0801.DXKX250549.

摘要

在复杂电磁对抗场景中，无线信道的时变噪声干扰和多域耦合效应使测试数据集的噪声分布特性与训练集的先验假设条件发生偏离，导致深度神经网络模型出现特征失配，进而引发基于静态信道假设的调制识别系统在跨域场景中的性能劣化。为应对这一挑战，提出了一种基于对比学习与实复域融合的动态噪声自动调制识别方法。在预训练阶段，借助BYOL（bootstrap your own latent）对比学习框架，构建实复域融合网络，强制模型通过自监督学习来深刻理解数据的内在结构，从而降低特征提取对噪声分布变化的敏感性，增强模型在不同信噪比条件下的泛化能力；在微调阶段，将短时傅里叶生成的复数时频谱输入实复域融合网络，提取信号的多维特征，使网络学习到与信道噪声无关的本质特征。综合以上两种策略，使模型能够有效地应对不同信噪比条件下的动态噪声干扰。实验结果表明，在待识别信号信噪比下降6 dB的情况下，所提方法相较于现有视觉变换器（vision transformer，ViT）等方法，识别准确率至少提升了23.98%。

Abstract

In complex electromagnetic countermeasure scenarios

the time-varying noise interference of wireless channels and multi-domain coupling effects cause the noise distribution characteristics of the test data set to deviate from the prior assumption conditions of the training set

leading to feature mismatch in deep neural network models and subsequently causing performance degradation of modulation recognition systems based on the static channel assumptions in the cross-domain scenarios. To address this challenge

a dynamic noise automatic modulation recognition method based on contrastive learning and real-complex domain fusion was proposed. In the pre-training stage

the bootstrap your own latent (BYOL) contrastive learning framework was utilized to construct a real-complex domain fusion network

forcing the model to deeply understand the intrinsic structure of the data through self-supervised learning

thereby reducing the sensitivity of feature extraction to the changes in noise distribution and enhancing the model

s generalization ability under different signal-to-noise ratio conditions. In the fine-tuning stage

the complex time-frequency spectrum generated by short-time Fourier transform was input into the real-complex domain fusion network to extract the multi-dimensional features of signal

enabling the network to learn the essential features that independent of channel noise. The combination of these two strategies enabled the model to effectively cope with the dynamic noise interference under different signal-to-noise ratio conditions. Experimental results show that when the signal-to-noise ratio of the signal to be recognized decreases by 6 dB

the proposed method achieves at least a 23.98% improvement in recognition accuracy compared with the existing methods

such as vision Transformer (ViT).

关键词

Keywords

references

刘京华 , 魏祥麟 , 范建华 , 等 . 基于时序深度残差收缩网络的混叠信号调制识别方法 [J ] . 电信科学 , 2024 , 40 ( 10 ): 27 - 38 .

Liu J H , Wei X L , Fan J H , et al . Mixed signal modulation recognition method based on temporal depth residual shrinkage network [J ] . Telecommunications Science , 2024 , 40 ( 10 ): 27 - 38 .

杨颖 . 无线信号智能识别方法研究 [D ] . 成都 : 电子科技大学 , 2023 .

Yang Y . Research on intelligent recognition method of wireless signal [D ] . Chengdu : University of Electronic Science and Technology of China , 2023 .

万进华 , 尚俊娜 , 张华娣 . 基于多通道轻量化的自动调制识别网络 [J ] . 电信科学 , 2025 , 41 ( 2 ): 41 - 56 .

Wan J H , Shang J N , Zhang H D . Automatic modulation recognition network based on multi-channel lightweight [J ] . Telecommunications Science , 2025 , 41 ( 2 ): 41 - 56 .

李欣然 , 李春国 , 明钰婷 , 等 . 脉冲噪声环境下的调制识别算法 [J ] . 信号处理 , 2024 , 40 ( 8 ): 1497 - 1507 .

Li X R , Li C G , Ming Y T , et al . Modulation recognition algorithm in impulsive noise environment [J ] . Journal of Signal Processing , 2024 , 40 ( 8 ): 1497 - 1507 .

Zhao D , Yang J , Liu H , et al . Specific emitter identification model based on improved BYOL self-supervised learning [J ] . Electronics , 2022 , 11 ( 21 ): 3485 .

谭凯文 , 高岩 , 石坚 , 等 . 基于DropBlock双模态混合神经网络的无线通信调制识别 [J ] . 电信科学 , 2022 , 38 ( 5 ): 75 - 86 .

Tan K W , Gao Y , Shi J , et al . Wireless communication modulation recognition based on DropBlock bimodal hybrid neural network [J ] . Telecommunications Science , 2022 , 38 ( 5 ): 75 - 86 .

杨发权 , 李赞 , 罗中良 . 基于一阶循环均值算法的VHF频段信号调制分类识别方法研究 [J ] . 电信科学 , 2014 , 30 ( 2 ): 76 - 81 .

Yang F Q , Li Z , Luo Z L . Research on modulation classification and recognition method for VHF band signals based on first-order cyclic mean algorithm [J ] . Telecommunications Science , 2014 , 30 ( 2 ): 76 - 81 .

Xu J , Luo C , Parr G , et al . A spatiotemporal multi-channel learning framework for automatic modulation recognition [J ] . IEEE Wireless Communications Letters , 2020 , 9 ( 10 ): 1629 - 1632 .

孙姝君 , 彭盛亮 , 姚育东 , 等 . 基于深度学习的调制识别综述 [J ] . 电信科学 , 2021 , 37 ( 5 ): 82 - 90 .

Sun S J , Peng S L , Yao Y D , et al . A survey of modulation recognition based on deep learning [J ] . Telecommunications Science , 2021 , 37 ( 5 ): 82 - 90 .

Pan Y , Yang S , Peng H , et al . Specific emitter identification based on deep residual networks [J ] . IEEE Access , 2019 , 7 : 54425 - 54434 .

Tang H , Li T , Qiu T , et al . Constant modulus algorithm for cochannel signal separation in non-Gaussian impulsive noise environments [J ] . Wireless Personal Communications , 2010 , 54 : 591 - 604 .

刘剑锋 , 于宏毅 , 杜剑平 , 等 . 基于领域自适应的动态噪声辐射源个体识别 [J ] . 信号处理 , 2021 , 37 ( 6 ): 1000 - 1007 .

Liu J F , Yu H Y , Du J P , et al . Individual identification of dynamic noise radiation source based on domain adaptation [J ] . Journal of Signal Processing , 2021 , 37 ( 6 ): 1000 - 1007 .

郭业才 , 姚文强 . 基于信噪比分类网络的调制信号分类识别算法 [J ] . 电子与信息学报 , 2022 , 44 ( 10 ): 3507 - 3515 .

Guo Y C , Yao W Q . Modulation signal classification and recognition algorithm based on signal-to-noise ratio classification network [J ] . Journal of Electronics & Information Technology , 2022 , 44 ( 10 ): 3507 - 3515 .

Xu L , Fu X , Wang Y , et al . Enhanced few-shot specific emitter identification via phase shift prediction and decoupling [C ] // Proceedings of the IEEE Transactions on Cognitive Communications and Networking . Piscataway : IEEE Press , 2025 : 145 - 155 .

Zhang Y , Zhang Q , Zhao H , et al . Multisource heterogeneous specific emitter identification using attention mechanism-based rff fusion method [J ] . IEEE Transactions on Information Forensics and Security , 2024 , 19 : 2639 - 2650 .

Zhang F , Luo C , Xu J , et al . an autoencoder-based I/Q channel interaction enhancement method for automatic modulation recognition [J ] . IEEE Transactions on Vehicular Technology , 2023 , 72 ( 7 ): 9620 - 9625

王子恒 , 张徐 , 高硕 , 等 . 一种复数域轻量化知识蒸馏驱动的调制识别模型 [J ] . 电信科学 , 2025 , 41 ( 8 ): 163 - 175 .

Wang Z H , Zhang X , Gao S , et al . A complex domain lightweight knowledge distillation driven modulation recognition model [J ] . Telecommunications Science , 2025 , 41 ( 8 ): 163 - 175 .

Shen G , Zhang J , Marshall A , et al . Toward length-versatile and noise-robust radio frequency fingerprint identification [J ] . IEEE Transactions on Information Forensics and Security , 2023 , 18 : 2355 - 2367 .

O’Shea T J , Corgan J , Clancy T C . Convolutional radio modulation recognition networks [C ] // Proceedings of the Engineering Applications of Neural Networks: 17th International Conference . Aberdeen, UK : Springer , 2016 : 213 - 226 .

Sathyanarayanan V , Gerstoft P , El Gamal A . Data centric approach to modulation classification [C ] // Proceedings of the 2023 15th International Conference on Communication Systems & Networks (COMSNETS) . Piscataway : IEEE Press , 2023 : 340 - 344 .

Liu C , Cai Z , Zhang B . Automatic modulation classification based on complex-valued convolutional neural network and semi-supervised learning [C ] // Proceedings of the 2023 IEEE 23rd International Conference on Communication Technology (ICCT) . Piscataway : IEEE Press , 2023 : 241 - 246 .

Ding L , Wang S , Wang F , et al . Specific emitter identification via convolutional neural networks [J ] . IEEE Communications Letters , 2018 , 22 ( 12 ): 2591 - 2594 .

Ding W , Guo X , Wang G . Radar-based human activity recognition using hybrid neural network model with multidomain fusion [J ] . IEEE Transactions on Aerospace and Electronic Systems , 2021 , 57 ( 5 ): 2889 - 2898 .

王天池 , 俞璐 , 赫德军 . 基于对抗的一致性正则半监督辐射源个体识别方法 [J ] . 计算机测量与控制 , 2023 , 31 ( 2 ): 204 - 209 .

Wang T C , Yu L , He D J . Adversarial-based consistency regularization semi-supervised specific emitter identification method [J ] . Computer Measurement & Control , 2023 , 31 ( 2 ): 204 - 209 .

Cheng R , Chen Q , Huang M . Automatic modulation recognition using deep CVCNN-LSTM architecture [J ] . Alexandria Engineering Journal , 2024 , 104 : 162 - 170 .

Zhang F , Luo C , Xu J , et al . An autoencoder-based I/Q channel interaction enhancement method for automatic modulation recognition [J ] . IEEE Transactions on Vehicular Technology , 2023 , 72 ( 7 ): 9620 - 9625 .

Liu C , Fu X , Wang Y , et al . Overcoming data limitations: a few-shot specific emitter identification method using self-supervised learning and adversarial augmentation [J ] . IEEE Transactions on Information Forensics and Security , 2024 , 19 : 500 - 513 .

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Cross-view contrastive learning and multi-channel hypergraph convolution for session-based recommendation

Node classification of complex network based on enhanced graph neural network and contrastive learning

Semantic embedding via joint loss function and composite contrastive learning

Dual branch guided contrastive learning for unsupervised pedestrian re-identification

Multi-pattern time-aware sequential recommendation with data augmentation

Related Author

FANG Chi

SUI Xinyi

WANG Ruiqin

REN Yubin

TAN Yanli

WANG Yu

XU Peiling

YANG Songhao

Related Institution

School of Information Engineering, Huzhou University

Department of Electronic Information, Shanxi Institute of Economics and Business

School of Information and Communication Engineering, North University of China

Department of Electronic Engineering, Taiyuan Institute of Technology

Power Information Technology Co., Ltd.

⁰