A Federated Siamese–Autoencoder Framework For Robust ECG-Based Biometric Authentication With Spoof Attack Detection
DOI:
https://doi.org/10.29284/kgj9c129Keywords:
ECG Biometrics, Federated Learning, Siamese Neural Network, Metric Learning, Spoof Detection, Autoencoder, Biometric Authentication, Deep Learning.Abstract
User authentication based on electrocardiogram (ECG) data has emerged as a convincing modality for continuous and unobtrusive biometric-based user authentication leveraging the liveness property and physiologically unique characteristics of ECG. Unfortunately, spoofing attacks make it a non-trivial problem in practice and thus will be a main target for research given their implications on data privacy. This study presents a unique two-stage deep learning framework that combines federated metric learning with reconstruction-based anomaly detection to provide secure ECG authentication. The first one is using a federated Siamese neural network to learn discriminative embedding on scattered clients without transmitting raw data, which protects the user but improves the generalization. The model employ triplet-based optimization to ensure high inter-subject separability and intra-subject compactness in the embedding space. The second stage involves training an autoencoder on only clean original ECG features for the purpose of creating a model of healthy physiological activity; spoofed or manipulated signals can then be identified by analysis of a reconstruction error. The proposed framework analyzes multi-domain ECG features derived from time, frequency, heart rate variability and wavelet domains to fully characterize cardiac signals. Experimental results show that the model is capable of achieving high authentication accuracy with low equal error rate and effectively discriminate between spoofed inputs under diverse attack situations. The combination of federated learning and reconstruction-based spoof detection leads to a scalable, privacy-presumably secure, and resilient approach for future biometric authentication systems.
References
1. Hazratifard M, Agrawal V, Gebali F, Elmiligi H, Mamun M. Ensemble Siamese Network (ESN) Using ECG Signals for Human Authentication in Smart Healthcare System. Sensors. 2023; 23(10): 4727. https://doi.org/10.3390/s23104727
2. Wang G, Shanker S, Nag A, Lian Y, John D. ECG Biometric Authentication Using Self-Supervised Learning for IoT Edge Sensors. arXiv preprint. 2024. https://arxiv.org/abs/2409.05627
3. Behrozi P, Shirkani B, Hazratifard M. Using ECG Signals in Siamese Networks for Authentication in Digital Healthcare Systems. Journal of Biomedical Research & Environmental Sciences. 2022; 3(11): 1367-1373. https://www.jelsciences.com/articles/jbres1605.php
4. Jayaprakash A, Patro KK, Samantray S, Plawiak P, Hammad M. A Deep Learning Technique for Biometric Authentication Using ECG Beat Template Matching. Information. 2023; 14(2): 2-16. https://www.researchgate.net/publication/367350712
5. Deshpande S, Getnet Y, Dargie W. Detection of Intelligent Tampering in Wireless Electrocardiogram Signals Using Hybrid Machine Learning. arXiv preprint. 2025. https://arxiv.org/abs/2507.06402
6. Rai DH, Kafley S. Lightweight MobileNetV1+GRU for ECG Biometric Authentication: Federated and Adversarial Evaluation. arXiv preprint. 2025. https://arxiv.org/abs/2509.20382
7. Kapsecker M, Jonas SM. Cross-device federated unsupervised learning for ECG anomaly detection using autoencoders. PLOS Digit Health. 2025; 4(4): e0000793. https://pmc.ncbi.nlm.nih.gov/articles/PMC11975069/
8. Ibtehaz N, Chowdhury MEH, Khandakar A, Kiranyaz S, Rahman MS, Tahir A, Qiblawey Y, Rahman T. EDITH: ECG Biometrics Aided by Deep Learning for Reliable Individual Authentication. arXiv preprint. 2021. https://arxiv.org/abs/2102.08026
9. Zhang Q, Zhou D, Zeng X. HeartID: A Multiresolution Convolutional Neural Network for ECG-Based Biometric Human Identification in Smart Health Applications. IEEE Access. 2017; 5: 11805-11816. https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7933065
10. Prakash AJ, Kiran KP, Hammad M, Tadeusiewicz R, Pławiak P. BAED: A secured biometric authentication system using ECG and deep learning. Biocybernetics and Biomedical Engineering. 2022; 42(4): 1081-1093. https://www.sciencedirect.com/science/article/abs/pii/S0208521622000778
11. Wang X, Cai W, Wang M. A novel approach for biometric recognition based on ECG feature vectors. Biomedical Signal Processing and Control. 2023; 86(A): 104922. https://www.sciencedirect.com/science/article/abs/pii/S1746809423003555
12. Lonbar SM, Beigi A, Bagheri N, Peris-Lopez P, Camara C. Deep learning based bio-metric authentication system using a high temporal/frequency resolution transform. Frontiers in Digital Health. 2024; 17(6): 1463713. https://pmc.ncbi.nlm.nih.gov/articles/PMC11685233/
13. Choi G, Ziyang G, Wu J, Esposito C, Choi C. Multi-modal biometrics based implicit driver identification System Using Multi-TF Images of ECG and EMG. Computers in Biology and Medicine. 2023; 159: 106851. https://www.sciencedirect.com/science/article/abs/pii/S0010482523003165
14. López JLC, Parra C, Forero G. An ECG Deep Learning user identification architecture using ECG sex recognition as a selective parameter. Informatics in Medicine Unlocked. 2024; 50: 101563. https://www.sciencedirect.com/science/article/pii/S2352914824001199
15. Alfi MA, Lopez PP, Camara C. Enhancing biometric identification using 12-lead ECG signals and graph convolutional networks. Frontiers in Digital Health. 2025; 7: 1547208. https://www.frontiersin.org/articles/10.3389/fdgth.2025.1547208
16. Li L, Chen C, Pan L, Tai Y, Zhang J, Xiang Y. Hiding Your Signals: A Security Analysis of PPG-Based Biometric Authentication. European Symposium on Research in Computer Security. 2023; 183-202. https://link.springer.com/chapter/10.1007/978-3-031-51479-1_10
17. Kim JS, Pan S. Fake Electrocardiogram Signal Analysis to Detect Spoofing Attacks in Biometrics. IEEE Sensors Journal. 2025; 25(14): 27221 - 27229. https://doi.org/10.1109/JSEN.2025.3577311
18. Garg A, Karimian N. ECG Biometric Spoofing Using Adversarial Machine Learning. IEEE International Conference on Consumer Electronics. 2021. https://par.nsf.gov/servlets/purl/10333140
19. Abdullahi SM, Sun S, Wang B, Wei N, Wang H. Biometric template attacks and recent protection mechanisms: A survey. Information Fusion. 2024; 103: 102144. https://www.sciencedirect.com/science/article/pii/S1566253523004608
20. Wang W, Zhang X, Du J, Zhao W. Robust continuous authentication via multi-channel photoplethysmography signals: A wearable wristband solution for uncontrolled environments. Computers & Security. 2025; 159: 104686. https://www.sciencedirect.com/science/article/abs/pii/S016740482500375X
21. Gimba UA, et al. Enhancing biometric authentication through multimodal approach combining face and fingerprint recognition using convolutional neural networks (CNN). Discover Computing. 2025; 28: 246. https://link.springer.com/article/10.1007/s10791-025-09775-z
22. Aguilar D, Martínez-Cruz A, Ramírez-Gutiérrez KA, Morales-Sandoval M. PPG-based biometric authentication: A review on architectures, datasets, attacks and security challenges. Computational and Structural Biotechnology Journal. 2025; 28: 511-528. https://pmc.ncbi.nlm.nih.gov/articles/PMC12664393/
23. Singh RP, Dash R, Mohapatra RK. Unveiling explainability in face anti-spoofing: Hybrid feature extraction with XAI-guided feature aggregation. Pattern Recognition. 2025; 169: 111905.
https://www.sciencedirect.com/science/article/pii/S0031320325005655
24. Dataset source: https://www.kaggle.com/datasets/bjoernjostein/ecgid-database
Downloads
Published
Issue
Section
License
This work is licensed under a Creative Commons Attribution 4.0 International License.
This work is licensed under a Creative Commons Attribution 4.0 International License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
