Abstract
A constraint-aware adversarially robust Internet of Things (IoT) traffic classification system with protocol validity, device behavior consistency, and manifold-aware training and evaluation is presented in this study. In realistic IoT communication semantics, resilience as a constrained min–max optimization problem allows adversarial perturbations. Comprehensive testing on sample IoT traffic datasets shows that baseline models achieve 95.1% accuracy under benign conditions but plummet following hostile attacks. The proposed defense reduces untargeted attack success rates to <18% while achieving 81.3% accuracy at ε=0.05 and 70.6% at ε=0.10. The proposed constraint-aware adversarial framework significantly enhances IoT traffic classification by achieving 97.4% accuracy and maintaining 90.6% robustness at ε=0.10, outperforming state-of-the-art methods. It reduces attack success rates to 11.2% (untargeted) and 7.9% (targeted) through protocol-compliant perturbations and manifold-aware learning. Additionally, the model achieves an efficient trade-off with 21.4 ms latency and 650 flows/sec throughput, making it suitable for real-time edge deployment. These results demonstrate improved robustness, realism, and deployability compared to existing approaches.
References
Tahaei H, Afifi F, Asemi A, Zaki F, Anuar NB. The rise of traffic classification in IoT networks: A survey. Journal of Network and Computer Applications, 2020, 154: 102538. doi:10.1016/j.jnca.2020.102538.
Somsuk K. Enhanced algorithm for recovering RSA plaintext when two modulus values share at least one common prime factor. Journal of Cyber Security and Mobility, 2025, 14(2): 433–456. doi:10.13052/jcsm2245-1439.1427.
Qiu H, Dong T, Zhang T, Lu J, Memmi G, Qiu M. Adversarial attacks against network intrusion detection in IoT systems. IEEE Internet of Things Journal, 2021, 8(13): 10327–10335. doi:10.1109/JIOT.2020.3048038.
Neto ECP, Dadkhah S, Ferreira R, Zohourian A, Lu R, Ghorbani AA. CICIoT2023: A real-time dataset and benchmark for large-scale attacks in IoT environments. Sensors, 2023, 23(13): 5941. doi:10.3390/s23135941.
Velichkovska B, Cholakoska A, Atanasovski V. Machine learning-based classification of IoT traffic. Radioengineering, 2023, 32(2): 256–263. doi:10.13164/re.2023.0256.
Fouad Y, Abdelaziz NE, Elshewey AM. IoT traffic parameter classification based on optimized BPSO for enabling green wireless networks. Engineering, Technology & Applied Science Research, 2024, 14(6): 18929–18934.
Pfeffer MA, Wong JKW, Ling SH. Trends and limitations in transformer-based BCI research. Applied Sciences, 2025, 15(20): 11150. doi:10.3390/app152011150.
Deng X, Pan Y, Fang H. Anomaly detection in smart grid behavior monitoring via federated learning: A privacy-preserving defense against cyber-physical attacks. Journal of Cyber Security and Mobility, 2025, 14(5): 1151–1172. doi:10.13052/jcsm2245-1439.1455.
Kim M, Joo S. Time-constrained adversarial defense in IoT edge devices through kernel tensor decomposition and multi-DNN scheduling. Sensors, 2022, 22(15): 5896. doi:10.3390/s22155896.
Wang K, Wang Z, Han D, Chen W, Yang J, Shi X, Yin X. BARS: Local robustness certification for deep learning-based traffic analysis systems. In: Proceedings of the Network and Distributed System Security Symposium (NDSS), 2023: 1–18.
Sharma S, Chen Z. A systematic study of adversarial attacks against network intrusion detection systems. Electronics, 2024, 13(24): 5030. doi:10.3390/electronics13245030.
Adeke JM, Liu G, Zhao J, Wu N, Bashir HM. Securing network traffic classification models against adversarial examples using derived variables. Future Internet, 2023, 15(12): 405. doi:10.3390/fi15120405.
Apruzzese G, Andreolini M, Ferretti L, Marchetti M, Colajanni M. Modeling realistic adversarial attacks against network intrusion detection systems. Digital Threats: Research and Practice, 2022, 3(3): Article 31. doi:10.1145/3469659.
Li M, Xu Y, Li N, Jin Z. Adversarial attacks on deep learning-based methods for network traffic classification. In: Proceedings of the IEEE International Conference on Trust, Security and Privacy in Computing and Communications (TrustCom), 2022: 1123–1128. doi:10.1109/TrustCom56396.2022.00154.
Areia J, Bispo IA, Santos L, Costa RL de C. IoMT-TrafficData: Dataset and tools for benchmarking intrusion detection in Internet of Medical Things. IEEE Access, 2024, 12: 1–15. doi:10.1109/ACCESS.2024.3437214.
Xiong WD, Luo KL, Li R. AIDTF: Adversarial training framework for network intrusion detection. Computers & Security, 2023, 128: 103141. doi:10.1016/j.cose.2023.103141.
Alshahrani E, Alghazzawi D, Alotaibi R, Rabie O. Adversarial attacks against supervised machine learning-based network intrusion detection systems. PLOS One, 2022, 17(10): e0275971. doi:10.1371/journal.pone.0275971.
Ci Z, Liu Z, Song Y, Qin F, Li Y, Zhao J. Multi-task class-aware adversarial training for remote sensing object detection robustness. Connection Science, 2025, 37(1). doi:10.1080/09540091.2025.2581373.
Yan J, Liu Z, Xie Y, Liang S, Liu L, Xu K. CertTA: Certified robustness made practical for learning-based traffic analysis. In: Proceedings of the 34th USENIX Security Symposium, 2025: 7349–7368.
Alhajjar E, Maxwell P, Bastian N. Adversarial machine learning in network intrusion detection systems. Expert Systems with Applications, 2021, 186: 115782. doi:10.1016/j.eswa.2021.115782.
Vitorino J, Dias T, Fonseca T et al. Constrained adversarial learning for automated software testing: A literature review. Discover Applied Sciences, 2025, 7: 547. doi:10.1007/s42452-025-07073-3.
Susilo B, Muis A, Sari RF. Intelligent intrusion detection system against various attacks based on a hybrid deep learning algorithm. Sensors, 2025, 25(2): 580. doi:10.3390/s25020580.
Quirumbay Yagual D, Fernández Iglesias D, Nóvoa FJ. A hybrid deep learning-based architecture for network traffic anomaly detection via EFMS-enhanced KMeans clustering and CNN-GRU models. Applied Sciences, 2025, 15(20): 10889. doi:10.3390/app152010889.
Abiha UE, Rehman A, Abbas A, Haider MA, Al-Yarimi FAM, Gul MU, Hassan SR. Improving adversarial resilience for anomaly detection in the heterogeneous Internet of Things through ensemble models. Future Generation Computer Systems, 2026, 178: 108299. doi:10.1016/j.future.2025.108299.
Dahanayaka T, Ginige Y, Huang Y, Jourjon G, Seneviratne S. Robust open-set classification for encrypted traffic fingerprinting. Computer Networks, 2023, 236: 109991. doi:10.1016/j.comnet.2023.109991.
Xia L, Wang G. Research on IoT data anonymization method based on multimodal adversarial generative networks and reinforcement learning collaborative optimization. Journal of Cyber Security and Mobility, 2025, 14(04): 799–822.
Heydari V, Nyarko K. Enhancing adversarial robustness in network intrusion detection: A novel adversarially trained neural network approach. Electronics, 2025, 14(16): 3249. doi:10.3390/electronics14163249.
Bi S, Wang J, Song J, Li P, Li L. Research on the intrusion detection model for power Internet of Things combining deep belief network and BiLSTM. Journal of Cyber Security and Mobility, 2025, 14(03): 653–672.
Baldoni S, Battisti F. Histogram-based network traffic representation for anomaly detection through PCA. Computer Networks, 2025, 265: 111276. doi:10.1016/j.comnet.2025.111276.
University of Padua. Network Anomaly Detection. Available: https://medialab.dei.unipd.it/research-areas/forensics-and-security/network-anomaly-detection/.
Dogra A. CIC-IoT Dataset 2023. Available: https://www.kaggle.com/datasets/akashdogra/cic-iot-2023.
University of New South Wales. The Bot-IoT Dataset. Available: https://research.unsw.edu.au/projects/bot-iot-dataset.
University of New South Wales. The TON_IoT Datasets. Available: https://research.unsw.edu.au/projects/toniot-datasets.
Moustafa N, Slay J. UNSW-NB15: A comprehensive data set for network intrusion detection systems (UNSW-NB15 network data set). In: 2015 Military Communications and Information Systems Conference (MilCIS), Canberra, ACT, Australia, 2015: 1–6. doi:10.1109/MilCIS.2015.7348942.
Dongxu Sun, Peng Li, Lina Pei, Chuhang Wei. Online Network Intrusion Detection Based on Self-Encoder Integration. In: Proceedings of the 3rd International Conference on Signal Processing, Computer Networks and Communications (SPCNC ’24). Association for Computing Machinery, New York, NY, USA, 2025: 246–250.
Toba AL, Kulkarni S, Khallouli W, Pennington T. Long-term traffic prediction using deep learning long short-term memory. Smart Cities, 2025, 8(4): 126. doi:10.3390/smartcities8040126.
Ma H, Zhang W, Zhang D et al. An IoT intrusion detection framework based on feature selection and large language models fine-tuning. Sci Rep, 2025, 15: 21158. doi:10.1038/s41598-025-08905-3.
Gautam S, Malhotra A, Dhurandher SK. A hybrid CNN-LSTM model for enhanced intrusion detection in Internet of Things environments. In: 2025 International Conference on Computational, Communication and Information Technology (ICCCIT), Indore, India, 2025, 781–786, doi:10.1109/ICCCIT62592.2025.10927909.
Xu S, Han J, Liu Y et al. Few-shot traffic classification based on autoencoder and deep graph convolutional networks. Sci Rep, 2025, 15: 8995. doi:10.1038/s41598-025-94240-6.
Madry Aleksander et al. Towards deep learning models resistant to adversarial attacks. In: 6th International Conference on Learning Representations, ICLR 2018 – Conference Track Proceedings, 2018. Available: https://openreview.net/forum?id=rJzIBfZAb.
Tchaleu BY, Ndjiongue AR, Leke CA. Generative adversarial networks: A comprehensive review and the way forward. SAIEE Africa Research Journal, 2025, 116(3): 101–124. doi:10.23919/SAIEE.2025.11090063.
Haiqi Zhang, Guangquan Lu, Mengmeng Zhan, Beixian Zhang. Semi-supervised classification of graph convolutional networks with Laplacian rank constraints. Neural Process Lett, 2022, 54(4): 2645–2656. doi:10.1007/s11063-020-10404-7.
Zhou, Z.-H. (2012). Ensemble Methods: Foundations and Algorithms. Chapman & Hall/CRC.
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
Copyright (c) 2026 Journal of Cyber Security and Mobility