A Diffusion Model Driven Approach for Cross-Time-Domain Incremental Specific Emitter Identification

ZHAO Dong-xing; LIU Hui; HUANG Ke-ju; YANG Jun-an

doi:10.12263/DZXB.20250843

您当前的位置：

首页 >

文章列表页 >

A Diffusion Model Driven Approach for Cross-Time-Domain Incremental Specific Emitter Identification

PAPERS | 更新时间：2026-04-24

- A Diffusion Model Driven Approach for Cross-Time-Domain Incremental Specific Emitter Identification
- ACTA ELECTRONICA SINICA Vol. 53, Issue 12, Pages: 4527-4540(2025)
- 作者机构：
  
  1.国防科技大学电子对抗学院电子对抗工程系，安徽合肥 230000
  2.安徽省电子制约技术重点实验室，安徽合肥 230000
- 作者简介：
- 基金信息：
  
  Independent Scientific Research Program of National University of Defense Technology(ZK24-44)
- DOI：10.12263/DZXB.20250843
  CLC： TN911.7;
- Received：25 September 2025，
  
  Accepted：22 December 2025，
  
  Published：25 December 2025
- 稿件说明：
移动端阅览
赵东兴, 刘辉, 黄科举, 等. 扩散模型驱动的跨时间域通信辐射源个体增量识别方法[J]. 电子学报, 2025, 53(12): 4527-4540.

ZHAO Dong-xing, LIU Hui, HUANG Ke-ju, et al. A Diffusion Model Driven Approach for Cross-Time-Domain Incremental Specific Emitter Identification[J]. Acta Electronica Sinica, 2025, 53(12): 4527-4540.
赵东兴, 刘辉, 黄科举, 等. 扩散模型驱动的跨时间域通信辐射源个体增量识别方法[J]. 电子学报, 2025, 53(12): 4527-4540. DOI：10.12263/DZXB.20250843

ZHAO Dong-xing, LIU Hui, HUANG Ke-ju, et al. A Diffusion Model Driven Approach for Cross-Time-Domain Incremental Specific Emitter Identification[J]. Acta Electronica Sinica, 2025, 53(12): 4527-4540. DOI：10.12263/DZXB.20250843

摘要

通信辐射源个体识别（Specific Emitter Identification，SEI）利用信号发射设备由于制造误差与器件老化等因素形成的细微硬件差异，在物理层完成身份区分与溯源.相较依赖协议与密钥的传统认证方式，它无需改动协议栈，对业务内容透明且部署成本低，已在频谱监管、无线安全、认知无线电与复杂电磁环境感知等场景展现重要价值.然而在真实无线环境中，跨时间与跨场景的信道变化会对射频指纹产生不稳定的调制与失真：多径衰落、频偏与相位噪声随时间漂移，使同一设备在不同时刻的信号表现差异显著，识别性能在目标域显著退化，成为落地应用的核心障碍.为缓解域间分布偏移，现有研究主要从迁移学习与领域自适应展开.迁移学习依赖目标域标注进行微调，虽可提升目标域性能，但易破坏源域已学知识并诱发灾难性遗忘；无监督领域自适应通过特征对齐、伪标签与熵最小化缩小分布差异，但因缺乏显式监督，提升幅度受限，且难以应对数据持续到来的在线场景.增量学习强调在不断接收新数据时兼顾“适应—记忆”的平衡，但多数方法仍需标注或额外存储，难以直接用于无线信号的无标注跨时间应用.生成式建模的发展为解决该问题提供了新契机.扩散模型以正向加噪、反向去噪的机制刻画复杂分布，具备建模“信道扰动—设备本征特征”叠加关系、从观测中恢复指纹的潜力，但既有工作多聚焦去噪或数据生成，尚未兼顾跨时间域识别与持续学习需求.为此，本文提出一种扩散模型驱动的跨时间域通信辐射源个体增量识别方法：在源域阶段以正向扩散显式建模信道扰动，在目标域通过反向扩散逐步恢复接近源域分布的判别性表示，以抑制特征漂移；在扩散网络中引入跨注意力，将个体信息注入去噪过程以增强类间可分性；进一步结合无监督增量学习，通过分布一致性与知识保持正则，仅依赖无标注目标样本实现持续自适应，并有效缓解遗忘.在WiSig数据集上，跨时间个体识别实验表明，所提方法在目标域识别准确率较典型领域自适应方法提升超过5个百分点，在源域性能保持方面较主流增量学习策略提升约10个百分点，验证了其信道还原与特征对齐能力，并体现出在动态信道条件下的鲁棒性与实用价值.

Abstract

Specific emitter identification (SEI) exploits subtle hardware discrepancies caused by manufacturing imperfections and device aging to perform transmitter identification and attribution at the physical layer. Compared with traditional authentication schemes that rely on protocols and cryptographic keys

SEI requires no modification to the protocol stack

is transparent to transmitted data

and incurs low deployment cost

making it valuable for applications such as spectrum regulation

wireless security

cognitive radio

and sensing in complex electromagnetic environments. However

in real-world wireless scenarios

time-varying and scene-dependent channel conditions introduce unstable modulation and distortion to radio-frequency fingerprints. Effects such as multipath fading

carrier frequency offset

and phase noise drift over time

causing the signals emitted by the same device to exhibit significant temporal variation. As a result

identification performance degrades markedly in the target domain

posing a major obstacle to practical deployment. To mitigate domain distribution shifts

existing studies mainly investigate transfer learning and domain adaptation approaches. Transfer learning relies on fine-tuning with labeled target-domain data and can improve target-domain performance

but it often disrupts previously learned source-domain knowledge and leads to catastrophic forgetting. Unsupervised domain adaptation reduces distribution discrepancies through feature alignment

pseudo labeling

and entropy minimization; however

due to the absence of explicit supervision

performance improvements are limited

and such methods struggle to handle continuously arriving data in online scenarios. Incremental learning emphasizes balancing adaptation to new data with the preservation of prior knowledge

yet most existing approaches still require labeled data or additional storage

making them difficult to apply directly to unlabeled cross-time SEI tasks. The advancement of generative modeling provides a new opportunity to address these challenges. Diffusion models characterize complex data distributions through forward noise injection and reverse denoising processes

and are well suited for modeling the superposition of channel perturbations and device-intrinsic features

enabling the recovery of radio-frequency fingerprints from distorted observations. Nevertheless

existing studies predominantly focus on denoising or data generation

and have not fully addressed cross-time identification and continual learning requirements. To this end

this paper proposes a diffusion-model-driven cross-time incremental SEI method. In the source domain

forward diffusion is employed to explicitly model channel perturbations

while in the target domain

reverse diffusion progressively restores discriminative representations that approximate the source-domain distribution

thereby suppressing feature drift. A cross-attention mechanism is incorporated into the diffusion network to inject emitter identity information during denoising

enhancing inter-class separability. Furthermore

an unsupervised incremental learning strategy is introduced

which achieves continual adaptation using only unlabeled target-domain samples through distribution consistency and knowledge-preservation regularization

effectively mitigating catastrophic forgetting. Cross-time identification experiments on the WiSig dataset demonstrate that the proposed method improves target-domain identification accuracy by more than 5 percentage points compared with representative domain adaptation methods

and enhances source-domain performance retention by approximately 10 percentage points relative to mainstream incremental learning strategies

validating its channel restoration capability

feature alignment effectiveness

and robustness under dynamic channel conditions.

关键词

Keywords

references

刘钊 , 马爽 , 张梦杰 , 等 . 多径条件下的雷达辐射源个体识别方法 [J ] . 电子学报 , 2023 , 51 ( 6 ): 1654 - 1665 .

LIU Z , MA S , ZHANG M J , et al . Radar specific emitter identification method under multipath conditions [J ] . Acta Electronica Sinica , 2023 , 51 ( 6 ): 1654 - 1665 . (in Chinese)

赵雅琴 , 谢丹 , 吴龙文 , 等 . 数字预失真下的辐射源个体识别技术 [J ] . 电子学报 , 2023 , 51 ( 11 ): 3331 - 3342 .

ZHAO Y Q , XIE D , WU L W , et al . Specific emitter identification under digital pre-distortion [J ] . Acta Electronica Sinica , 2023 , 51 ( 11 ): 3331 - 3342 . (in Chinese)

袁晔 , 汤春阳 , 张博轩 , 等 . 融合时频特征的通信辐射源个体识别方法 [J ] . 电子测量技术 , 2025 , 48 ( 1 ): 129 - 136 .

YUAN Y , TANG C Y , ZHANG B X , et al . Individual identification method for communication radiation sources by integrating time-frequency characteristics [J ] . Electronic Measurement Technology , 2025 , 48 ( 1 ): 129 - 136 . (in Chinese)

李奇真 , 刘佳旭 , 梁先明 , 等 . 基于深度学习的跨域辐射源个体识别综述 [J ] . 电讯技术 , 2024 , 64 ( 7 ): 1163 - 1174 .

LI Q Z , LIU J X , LIANG X M , et al . Cross-domain specific emitter identification based on deep learning: A comprehensive survey [J ] . Telecommunication Engineering , 2024 , 64 ( 7 ): 1163 - 1174 . (in Chinese)

孙丽婷 , 柳征 , 黄知涛 . 接收域分离的跨接收系统通用性辐射源指纹识别 [J ] . 电子与信息学报 , 2024 , 46 ( 10 ): 3966 - 3978 .

SUN L T , LIU Z , HUANG Z T . Universal radio frequency fingerprinting across receiving systems using receiving domain separation [J ] . Journal of Electronics & Information Technology , 2024 , 46 ( 10 ): 3966 - 3978 . (in Chinese)

徐宇恒 , 程嗣怡 , 董晓璇 , 等 . 基于DBN特征提取的雷达辐射源个体识别 [J ] . 空军工程大学学报(自然科学版) , 2019 , 20 ( 6 ): 91 - 96, 108 .

XU Y H , GHENG S Y , DONG X X , et al . Radar specific emitter identification based on DBN feature extraction [J ] . Journal of Air Force Engineering University (Natural Science Edition) , 2019 , 20 ( 6 ): 91 - 96, 108 . (in Chinese)

王美玉 . 数据受限下基于迁移学习的特定辐射源精准识别方法研究 [D ] . 哈尔滨 : 哈尔滨工程大学 , 2023 .

WANG M Y . Research on Accurate Identification Method of Specific Emitter Based on Transfer Learning under Data Constraints [D ] . Harbin : Harbin Engineering University , 2023 . (in Chinese)

刘振 , 陈阿磊 , 李世飞 , 等 . 基于迁移学习的辐射源个体识别分类方法 [J ] . 舰船电子对抗 , 2023 , 46 ( 2 ): 53 - 60, 78 .

LIU Z , CHEN A L , LI S F , et al . Specific emitter identification and classification method based on transfer learning [J ] . Shipboard Electronic Countermeasure , 2023 , 46 ( 2 ): 53 - 60, 78 . (in Chinese)

XIAO Z L , XIE Y H , LI Q , et al . Cross-domain specific emitter identification based on domain-specific classifier [J ] . IEEE Internet of Things Journal , 2025 , 12 ( 14 ): 27660 - 27670 .

LIANG J , HU D , FENG J . Do we really need to access the source data source hypothesis transfer for unsupervised domain adaptation [C ] // International Conference on Machine Learning . Cambridge : PMLR , 2020 : 6028 - 6039 .

VARGAS HAKIM G A , OSOWIECHI D , NOORI M , et al . CLIPArTT: Adaptation of CLIP to new domains at test time [C ] // 2025 IEEE/CVF Winter Conference on Applications of Computer Vision . Piscataway : IEEE , 2025 : 7092 - 7101 .

刘展阳 , 刘进锋 . 基于知识蒸馏的不存储旧数据的类增量学习 [J ] . 计算机应用 , 2024 , 44 ( S2 ): 12 - 17 .

LIU Z Y , LIU J F . Data-free class incremental learning based on knowledge distillation [J ] . Journal of Computer Applications , 2024 , 44 ( S2 ): 12 - 17 . (in Chinese)

WANG S Y , LUO X Q . Incremental learning method for lung nodule detection based on EWC and feature distillation [C ] // International Conference on Biomedical and Intelligent Systems (IC-BIS 2022) . Chengdu : SPIE , 2022 : 868 - 874 .

DENG Z W , LI C , SONG R C , et al . Centroid-guided domain incremental learning for EEG-based seizure prediction [J ] . IEEE Transactions on Instrumentation and Measurement , 2024 , 73 : 2500613 .

段可欣 , 闫文君 , 刘凯 , 等 . 基于动态权重模型的数据不平衡SEI方法 [J ] . 太赫兹科学与电子信息学报 , 2024 , 22 ( 2 ): 142 - 151, 159 .

DUAN K X , YAN W J , LIU K , et al . Data imbalance SEI method based on dynamic weight model [J ] . Journal of Terahertz Science and Electronic Information Technology , 2024 , 22 ( 2 ): 142 - 151, 159 . (in Chinese)

GUO H , GU Y N , ZHAO J Y , et al . Multiscale attention denoising diffusion probability model for multilingual handwriting character recognition [J ] . The Visual Computer , 2025 , 41 ( 11 ): 8467 - 8485 .

ZHANG C , YANG W X , LI X , et al . MMGInpainting: Multi-modality guided image inpainting based on diffusion models [J ] . IEEE Transactions on Multimedia , 2024 , 26 : 8811 - 8823 .

YING Z L , KE W Y , ZHAI Y K , et al . DiffuSAR: Frequency domain-aware diffusion model for SAR image generation [J ] . IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing , 2025 , 18 : 11851 - 11866 .

HO J , JAIN A , ABBEEL P . Denoising diffusion probabilistic models [C ] // Proceedings of the 34th International Conference on Neural Information Processing Systems . New York : ACM , 2020 : 6840 - 6851 .

WANG W Y , DOU Z , FU J Z , et al . A diffusion model-based open set identification method for specific emitters [C ] // 2023 IEEE/CIC International Conference on Communications in China . Piscataway : IEEE , 2023 : 1 - 5 .

ZHA H R , WANG W Y , YU S , et al . SEI-DM: Crafting revolutionary RF signals in low-resource settings via diffusion model [C ] // 2023 IEEE 23rd International Conference on Communication Technology . Piscataway : IEEE , 2024 : 890 - 895 .

ROMBACH R , BLATTMANN A , LORENZ D , et al . High-resolution image synthesis with latent diffusion models [C ] // 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition . Piscataway : IEEE , 2022 : 10674 - 10685 .

HANNA S , KARUNARATNE S , CABRIC D . WiSig: A large-scale WiFi signal dataset for receiver and channel agnostic RF fingerprinting [J ] . IEEE Access , 2022 , 10 : 22808 - 22818 .

CHEN J W , YU L , CHEN Y F , et al . Prototype-driven unsupervised domain adaptation for specific emitter identification [J ] . IEEE Internet of Things Journal , 2025 , 12 ( 1 ): 637 - 649 .

ZHANG X L , LI T Y , GONG P , et al . Variable-modulation specific emitter identification with domain adaptation [J ] . IEEE Transactions on Information Forensics and Security , 2023 , 18 : 380 - 395 .

MA Y , CHAI L , TU S , et al . Exploring relational knowledge for source-free domain adaptation [J ] . IEEE Transactions on Circuits and Systems for Video Technology , 2025 , 35 ( 2 ): 1825 - 1839 .

YAN L , ZHENG W B , LI K L . Source-free domain adaptive person search [J ] . Pattern Recognition , 2025 , 161 : 111317 .

杨宁 , 张邦宁 , 郭道省 . 基于元学习的小样本类增量辐射源个体识别方法 [J ] . 陆军工程大学学报 , 2025 , 4 ( 4 ): 46 - 52 .

YANG N , ZHANG B N , GUO D X . Few-shot class-incremental specific emitter identification method based on meta-learning [J ] . Journal of Army Engineering University of PLA , 2025 , 4 ( 4 ): 46 - 52 . (in Chinese)

LI D Z , CHEN X W , HONG S H , et al . FSCIL-SEI: Few-shot class-incremental learning approach for specific emitter identification [J ] . IEEE Transactions on Instrumentation and Measurement , 2025 , 74 : 2504914 .

张立民 , 谭凯文 , 闫文君 , 等 . 基于持续学习和联合特征提取的特定辐射源识别 [J ] . 电子与信息学报 , 2023 , 45 ( 1 ): 308 - 316 .

ZHANG L M , TAN K W , YAN W J , et al . Specific emitter identification based on continuous learning and joint feature extraction [J ] . Journal of Electronics & Information Technology , 2023 , 45 ( 1 ): 308 - 316 . (in Chinese)

HUA M Y , ZHANG Y B , ZHANG Q Y , et al . KG-IBL: Knowledge graph driven incremental broad learning for few-shot specific emitter identification [J ] . IEEE Transactions on Information Forensics and Security , 2024 , 19 : 10016 - 10028 .

李健 , 刘欢 , 李艳秋 , 等 . 基于THGS算法优化ResNet-18模型的图像识别 [J ] . 吉林大学学报(工学版) , 2025 , 55 ( 5 ): 1629 - 1637 .

LI J , LIU H , LI Y Q , et al . Image recognition research on optimizing ResNet-18 model based on THGS algorithm [J ] . Journal of Jilin University (Engineering and Technology Edition) , 2025 , 55 ( 5 ): 1629 - 1637 . (in Chinese)

ZHAO D X , HUANG K J , LIU H , et al . Innovative specific emitter identification with self-supervised and incremental learning [J ] . IEEE Transactions on Cognitive Communications and Networking , 2026 , 12 : 2724 - 2736 .

SHEN J H , AIN Q T , LIU Y H , et al . GTAT: Empowering graph neural networks with cross attention [J ] . Scientific Reports , 2025 , 15 : 4760 .

MAATEN L , HINTON G . Visualizing data using t-SNE [J ] . Journal of Machine Learning Research , 2008 , 9 ( 86 ): 2579 - 2605 .

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Towards Incremental Object Detection via Hierarchical Proposal and Decoupled Supervision

Progressive Image Synthesis Method Based on Diffusion-Mamba and Scale-Invariant Loss

Real-Time Gesture Recognition Algorithm Based on LoRa Signals

A Method for Continuous Detection and Classification of Malicious Network Traffic Based on Double-Layer Model and Distribution of Indexes

Related Author

LIANG Jia-wei

LIANG Si-yuan

CHEN Ruo-yu

LIU Kuan-rong

HUANG Jian-jie

CAO Xiao-chun

LI Hao

HAO Wen-ning

Related Institution

School of Cyber Science and Technology, Sun Yat-sen University

College of Computing and Data Science, Nanyang Technological University

Institute of Information Engineering, Chinese Academy of Sciences

School of Cyber Security, University of Chinese Academy of Sciences

College of Command and Control Engineering, Army Engineering University of PLA

⁰