Unsupervised Zero-Day Intrusion Detection in IoT Networks using Cycle-Consistent Adversarial Networks
DOI:
https://doi.org/10.63282/3050-9416.IJAIBDCMS-V7I2P123Keywords:
Internet of Things (IoT), Zero-Day Detection, Cycle-Consistent Adversarial Networks (CycleGAN), Unsupervised Learning, Anomaly Detection, Network SecurityAbstract
The rapid growth of the Internet of Things (IoT) has created new vulnerabilities in global networks, as traditional signature-based intrusion detection systems (IDS) find it difficult to detect new "zero-day" threats. This study explores the use of Cycle-Consistent Adversarial Networks (CycleGAN) for unsupervised anomaly detection. By employing unpaired domain translation, CycleGAN models the statistical patterns of normal network traffic and detects intrusions through reconstruction-error analysis. The report addresses key challenges, including data imbalance, resource limitations in edge computing, and the emergence of polymorphic attacks. Experimental results show that CycleGAN-based frameworks achieve higher detection rates on benchmark datasets while keeping false positives low.
References
1. Surepalli, Sirisha & Sameera, Nerella. (Feb 2026). Unsupervised Intrusion Detection System for Zero-Day Attack Detection Using Machine Learning and Deep Learning. 10.1007/978-3-032-14197-2_39.
2. R. S. et al., "Distributed Preprocessing and Adversarial Training for Robust IoT IDS," arXiv:2507.19739v1, 2025. [Online]. Available: https://arxiv.org/html/2507.19739v1
3. Z. Dehghanian, S. Saravani, M. Amirmazlaghani, and M. Rahmati, "Anomaly Detection Using Complete Cycle Consistent Generative Adversarial Network," International Journal of Neural Systems, vol. 35, no. 02, 2550004, 2025. [Online]. Available: https://doi.org/10.1142/S0129065725500042
4. K. Nitrat, N. Suetrong and N. Promsuk, "Zero-Day Attack Detection in IoT Networks Using a Residual Vision Transformer-Based Approach With Zero-Shot Learning," in IEEE Open Journal of the Communications Society, vol. 6, pp. 7405-7423, 2025, doi: 10.1109/OJCOMS.2025.3604826.
5. Ioannou I, Vassiliou V. Generative Adversarial Networks for Energy-Aware IoT Intrusion Detection: Comprehensive Benchmark Analysis of GAN Architectures with Accuracy-per-Joule Evaluation. Sensors (Basel). 2026 Jan 23;26(3):757. doi: 10.3390/s26030757. PMID: 41682273; PMCID: PMC12899382.
6. Fang, M., Wang, Y., Yang, L., Wu, H., Yin, Z., Liu, X., Xie, Z., & Kong, Z. (2024). Reinventing Web Security: An Enhanced Cycle-Consistent Generative Adversarial Network Approach to Intrusion Detection. Electronics, 13(9), 1711. https://doi.org/10.3390/electronics13091711
7. Alshehri, Mohammed & Saidani, Oumaima & Al Malwi, Wajdan & Asiri, Fatima & Latif, Shahid & Khattak, Aizaz & Ahmad, Jawad. (2025). A Hybrid Wasserstein GAN and Autoencoder Model for Robust Intrusion Detection in IoT. Computer Modeling in Engineering & Sciences. 143. 3899-3920. 10.32604/cmes.2025.064874.
8. F. S. Atedjio, J. -P. Lienou, F. F. Nelson, S. S. Shetty and C. A. Kamhoua, "CycleGAN-Gradient Penalty for Enhancing Android Adversarial Malware Detection in Gray Box Setting," in IEEE Access, vol. 12, pp. 162685-162696, 2024, doi: 10.1109/ACCESS.2024.3486734.
9. Sridharan, S., Patil, S., Shobha, T., Pai, P. (2025). Hybrid machine learning–based intrusion detection for zero-day attack prevention in digital education networks. International Journal of Safety and Security Engineering, Vol. 15, No. 8, pp.1703-1713. https://doi.org/10.18280/ijsse.150815
10. Hashim, Khalid & Mohd, Yusnani & Shahbudin, Shahrani. (2025). Mitigating Zero-Day Vulnerabilities in IIoT Systems: Challenges and Advances in AI-Powered Intrusion Detection Systems. Mesopotamian Journal of CyberSecurity. 5. 1184-1198. 10.58496//MJCS/2025/063.
11. Li, E., Shang, Z., Gungor, O., & Rosing, T. (2025). SAFE: Self-Supervised Anomaly Detection Framework for Intrusion Detection. ArXiv. https://arxiv.org/abs/2502.07119
12. Zahra, Fatima & Bostanci, Yavuz & Soyturk, Mujdat. (2024). Unsupervised Machine Learning for Anomaly Detection in Wi-Fi Based IoT Networks. 10.1109/ICCSPA61559.2024.10794232.
13. Dehghanian, Z., Saravani, S., Amirmazlaghani, M., & Rahmati, M. (2023). Spot The Odd One Out: Regularized Complete Cycle Consistent Anomaly Detector GAN. ArXiv. https://arxiv.org/abs/2304.07769
14. Jiang, YaPing & Zhang, ZhengHe & Ge, YangTao. (2024). CycleGAN-based intrusion detection data augmentation model. 350. 10.1117/12.3031413.
15. Ioannou I, Vassiliou V. Generative Adversarial Networks for Energy-Aware IoT Intrusion Detection: Comprehensive Benchmark Analysis of GAN Architectures with Accuracy-per-Joule Evaluation. Sensors (Basel). 2026 Jan 23;26(3):757. doi: 10.3390/s26030757. PMID: 41682273; PMCID: PMC12899382.
16. Jamoos, M.; Mora, A.M.; AlKhanafseh, M.; Surakhi, O. A New Data-Balancing Approach Based on Generative Adversarial Network for Network Intrusion Detection System. Electronics 2023, 12, 2851, doi:10.3390/electronics12132851.
17. Allagi, S., Pawan, T., & Leong, W. Y. (2025). Enhanced Intrusion Detection Using Conditional-Tabular-Generative-Adversarial-Network-Augmented Data and a Convolutional Neural Network: A Robust Approach to Addressing Imbalanced Cybersecurity Datasets. Mathematics, 13(12), 1923. https://doi.org/10.3390/math13121923
18. Gao, Ziyuan. (2025). Anomaly Detection for Enhancing IoT Device Security Using Machine Learning: A Comparative Study of Four Lightweight Models Based on the IoT-23 Dataset. ITM Web of Conferences. 80. 01027. 10.1051/itmconf/20258001027.
19. Wakili, A., & Bakkali, S. (Mar 2026). ZeroDefense: An adaptive hybrid fusion-based intrusion detection system for zero-day threat detection in IoT networks. Journal of Electronic Science and Technology, 24, Article 100345. https://doi.org/10.1016/j.jnlest.2026.100345
20. Lenort, M., Szygula, J., Marek, D., Marszalek, K., & Domanski, A. (2025). Comparative analysis of generator architectures in CycleGAN for image style transfer. 2025 IEEE International Conference on Big Data (BigData), 3979–3987. https://doi.org/10.1109/BigData66926.2025.11401638
21. D. P. Kavadi et al., "Design of an Integrated Model Combining CycleGAN, PPO, and Vision Transformer for Adaptive Scene Rendering in the Metaverse," in IEEE Access, vol. 13, pp. 21117-21138, 2025, doi: 10.1109/ACCESS.2025.3532327.
