基于多模态特征融合的通信干扰识别方法

蒙童远; 胡邓华; 石娟; 孙际哲; 王振涛

doi:10.12382/bgxb.2026.0301

您当前的位置：

首页 >

文章列表页 >

基于多模态特征融合的通信干扰识别方法

更新时间：2026-08-12

- 基于多模态特征融合的通信干扰识别方法
- Multimodal Feature Fusion-Based Method for Communication Interference Recognition
- 兵工学报 2026年
- 作者机构：
  
  空军工程大学防空反导学院, 陕西西安710051
- 作者简介：
  
  邮箱：hudenghua@qq.com
  邮箱：shijuan@mail.nwpu.edu.cn
- 基金信息：
  
  青年科技英才项目(2024)
- DOI：10.12382/bgxb.2026.0301
  中图分类号： TJ301
- 收稿：2026-04-02，
  
  修回：2026-06-02，
  
  录用：2026-06-11，
- 稿件说明：
移动端阅览
蒙童远, 胡邓华, 石娟, 等. 基于多模态特征融合的通信干扰识别方法[J/OL]. 兵工学报, 2026.

MENG Tongyuan, HU Denghua, SHI Juan, et al. Multimodal Feature Fusion-Based Method for Communication Interference Recognition[J/OL]. Acta Armamentarii, 2026.
蒙童远, 胡邓华, 石娟, 等. 基于多模态特征融合的通信干扰识别方法[J/OL]. 兵工学报, 2026. DOI： 10.12382/bgxb.2026.0301.

MENG Tongyuan, HU Denghua, SHI Juan, et al. Multimodal Feature Fusion-Based Method for Communication Interference Recognition[J/OL]. Acta Armamentarii, 2026. DOI： 10.12382/bgxb.2026.0301.

摘要

针对复杂电磁环境下通信干扰信号类型识别准确率低、单一特征难以全面表征信号的问题，提出了一种基于多分支注意力机制融合网络(multi-branch attention fusion network

MBAFNet)的干扰识别方法。该网络同时处理循环谱图、星座图、功率谱图和频域序列四种模态数据，针对不同模态特性选用不同网络提取深层特征；引入可学习加权机制自适应调整各模态贡献度，并通过自注意力模块对融合特征进行增强，充分挖掘特征维度间的内在依赖关系。实验结果表明，该方法在9类干扰信号识别任务中取得了优异的性能，在JNR ≥ -6 dB条件下识别准确率达到98%以上，识别性能优于传统方法。该网络方法充分融合了干扰信号的多域互补信息，具有较强的鲁棒性和泛化能力，为复杂电磁环境下通信干扰信号的精确识别提供了有效方法。

Abstract

Addressing the challenges of low accuracy in identifying communication jamming signal types and the limitations of single features in comprehensively characterizing signals under complex electromagnetic conditions

this study proposes a multi-branch attention fusion network named MBAFNet. This network simultaneously processes four modalities of data: cyclic spectrum

constellation diagram

power spectrum

and frequency-domain sequence. It employs different networks to extract deep features tailored to the characteristics of each modality. A learnable weighting mechanism is introduced to adaptively adjust the contribution of each modality

while self-attention module enhances the fused features to fully exploit the inherent dependencies among feature dimensions. Experimental results demonstrate that the proposed method achieves recognition accuracy of over 98% at JNRs not less than -6 dB and significantly outperforming traditional approaches. By fully integrating multi-domain complementary information from interference signals

the network exhibits strong robustness and generalization capabilities

providing an effective solution for precise interference signal identification in complex electromagnetic environments.

关键词

Keywords

references

姚富强 . 通信抗干扰工程与实践 [M ] . 第3版 . 北京 : 电子工业出版社 , 2025 : 47 .

YAO F Q . Communication Anti-jamming Engineering and Practice [M ] . 3rd ed . Beijing : Publishing House of Electronics Industry , 2025 : 47 . (in Chinese)

CALATRAVA H , TANG S , CLOSAS P . Advances in anti-deception jamming strategies for radar systems: a sur-vey [J ] . IEEE Aerospace and Electronic Systems Magazi-ne , 2026 , 41 ( 4 ): 66 - 93 .

WU Z L , ZHAO Y L , YIN Z D , et al . Jamming signals classification using convolutional neural network [C ] // 2017 IEEE International Symposium on Signal Pro-cessing and Information Technology (ISSPIT) . 2017 : 062 - 067 .

WU X J , ZHOU Y B , WU D L , et al . Improved deep residual shrinkage network for intelligent interference recognition with unknown interference [J ] . Sensors , 2023 , 23 ( 18 ): 7909 .

TANG Y , ZHAO Z J , CHEN J , et al . Open world recognition of communication jamming signals [J ] . China Communications , 2023 , 20 ( 6 ): 199 - 214 .

QU Q Z , WEI S J , LIU S , et al . JRNet: jamming recognition networks for radar compound suppression jamming signals [J ] . IEEE Transactions on Vehicular Technology , 2020 , 69 ( 12 ): 15035 - 15045 .

WANG P Y , CHENG Y F , PENG Q H , et al . Learning dynamic computing resource allocation in convolutional neural networks for wireless interference identification [J ] . IEEE Transactions on Vehicular Technology , 2023 , 72 ( 7 ): 8771 - 8782 .

LUO Z Y , CAO Y H , YEO T S , et al . SSJR-net: semi-supervised few-shot radar jamming recognition net-work [J ] . IEEE Transactions on Aerospace and Electronic Systems , 2025 , 61 ( 5 ): 12935 - 12950 .

陈炽根 , 彭启航 , 徐俊如 , 等 . 基于多任务学习的干扰认知方法 [J ] . 电子科技大学学报 , 2025 , 54 ( 6 ): 896 - 905 .

CHEN C G , PENG Q H , XU J R , et al . A jamming cognition approach based on multi-task learning [J ] . Journal of University of Electronic Science and Technology of China , 2025 , 54 ( 6 ): 896 - 905 . (in Chinese)

WANG D Y , DU L S Y , LI Z , et al . WPMNet: weight pruning and parameter quantization assisted multi-scale lightweight network for interference recognition [J ] . IEEE Transactions on Cognitive Communications and Net-working , 2026 , 12 : 1313 - 1326 .

WANG P Y , HUANG X , ZHANG H J , et al . Intelligent wireless interference recognition: challenges, opportu-nities, and future directions [J ] . IEEE Wireless Communi-cations , 2025 , 32 ( 6 ): 209 - 215 .

LIU M Q , LIU Z L , LU W D , et al . Distributed Few-Shot Learning for Intelligent Recognition of Communication Jamming [J ] . IEEE Journal of Selected Topics in Signal Processing , 2022 , 16 ( 3 ): 395 - 405 .

杨渡佳 , 张欣然 , 宋常建 , 等 . 基于多模态异构小波分解神经网络的干扰样式识别 [J ] . 通信学报 , 2025 , 46 ( 10 ): 150 - 162 .

YANG D J , ZHANG X R , SONG C J , et al . Jamming style recognition based on multimodal heterogeneous wavelet decomposition neural networks [J ] . Journal on Communications , 2025 , 46 ( 10 ): 150 - 162 . (in Chinese)

YAN H L , WANG G P , GAO F F , et al . Dynamic convolutional neural network for wireless interference identification [J ] . IEEE Communications Letters , 2024 , 28 ( 1 ): 83 - 87 .

WANG P Y , MA K , BAI Y S , et al . Wireless interference recognition with multimodal learning [J ] . IEEE Transac-tions on Wireless Communications , 2024 , 23 ( 12 ): 18576 - 18591 .

王珣 , 吴志强 , 刘明骞 . 基于ViT网络模型自注意力机制的干扰类型分布式识别方法 [J ] . 兵工学报 , 2025 , 46 .

WANG X , WU Z Q , LIU M Q . A Distributed Recognition Method for Interference Types based on the Self-attention Mechanism of ViT Network Model [J ] . ACTA ARMA-MENTARII , 2025 , 46 . (in Chinese)

田弘宇 , 葛松虎 , 郭宇 , 等 . 低干噪比下多特征融合的通信干扰识别方法 [J ] . 系统工程与电子技术 , 2025 .

TIAN H Y , GE S H , GUO Y , et al . Interference recognition method based on multi-feature fusion under low INR conditions [J ] . Systems Engineering and Electronics , 2025 . (in Chinese)

ZHOU S C , ZHANG Y , ZHU L D , et al . Wireless interference recognition based on multimodal feature fusion [J ] . IEEE Communications Letters , 2026 , 30 : 477 - 481 .

ZHANG Z H , AN J P , YE N , et al . Multi-domain fusion for high-sensitivity interference identification: a lightweight model using multi-scale extraction and compression [J ] . IEEE Transactions on Cognitive Com-munications and Networking , 2026 , 12 : 2675 - 2689 .

LIANG W X , TADESSE G A , HO D , et al . Advances, challenges and opportunities in creating data for trust-worthy AI [J ] . Nature Machine Intelligence , 2022 , 4 ( 8 ): 669 - 677 .

LI J Y , MOHAMED A , ZWEIG G , et al . LSTM time and frequency recurrence for automatic speech recogni-tion [C ] // 2015 IEEE Workshop on Automatic Speech Recognition and Understanding (ASRU) . 2015 : 187 - 191 .

浏览量

下载量

CNKI被引量

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于峰度值估计的无线信号频谱感知方法

混合架构的多尺度特征交互去雾算法

基于卷积神经网络与支持向量机的适配器落点预测方法

基于光照感知和特征增强的可见光-热红外图像语义分割

基于卷积神经网络和时空特征的烟幕视频检测和参数提取