Detecting AI-Generated Network Traffic Using Transformer–MLP Ensemble
Source abstract
Abstract
The rapid growth of generative artificial intelligence (AI) has enabled diverse applications but also introduced new attack techniques. Similar to deepfake media, generative AI can be exploited to create AI-generated traffic that evades existing intrusion detection systems (IDSs). This paper proposes a Dual Detection System to detect such synthetic network traffic in the Message Queuing Telemetry Transport (MQTT) protocol widely used in Internet of Things (IoT) environments. The system operates in two stages: (i) primary filtering with a Long Short-Term Memory (LSTM) model to detect malicious traffic, and (ii) secondary verification with a Transformer–MLP ensemble to identify AI-generated traffic. Experimental results show that the proposed method achieves an average accuracy of 99.1 ± 0.6% across different traffic types (normal, malicious, and AI-generated), with nearly 100% detection of synthetic traffic. These findings demonstrate that the proposed dual detection system effectively overcomes the limitations of single-model approaches and significantly enhances detection performance.
Source: crossref · Quality: baseline
Research context
OpenAlex topics: Network Security and Intrusion Detection · Internet Traffic Analysis and Secure E-voting · Anomaly Detection Techniques and Applications
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
Affiliations
Department of Software Engineering, Jeonbuk National University, Jeonju 54896, Republic of Korea
Funding
Basic Science Research Program
References (23)
Show source-provided references
1. (2025, August 22). Electricity AMI. Available online: https://www.aitimes.com/news/articleView.html?idxno=141421.
2. (2025, August 22). Gas AMI. Available online: https://www.gasnews.com/news/articleView.html?idxno=104555.
3. (2025, August 22). Water AMI. Available online: https://www.boannews.com/media/view.asp?idx=85538.
4. (2025, August 22). Global ICT Research. Available online: https://www.globalict.kr/product/product_list.do?menuCode=040200&knwldNo=143735.
5. (2025, August 22). MQTT. Available online: https://mqtt.org.
6. (2025, August 22). IOT Security Threat Report. Available online: https://www.msit.go.kr/bbs/view.do?sCode=user&nttSeqNo=3185279&pageIndex=&searchTxt=&searchOpt=ALL&bbsSeqNo=94&mId=307&mPid=208.
7. (2025, August 22). Three Open Source MQTT Message Brokers Found Vulnerable Against a DoS. Available online: https://www.secureblink.com/cyber-security-news/three-open-source-mqtt-message-brokers-found-vulnerable-against-a-dos-cyrc-alerted.
8. (2025, August 22). 32,000 Smart Homes Can Be Easily Hacked Due to Misconfigured MQTT Servers. Available online: https://www.csoonline.com/article/566079/32000-smart-homes-can-be-easily-hacked-due-to-misconfigured-mqtt-servers.html.
9. IDS-INT: Intrusion Detection System Using Transformer-Based Transfer Learning for Imbalanced Network Traffic10.1016/j.dcan.2023.03.008 ↗
10. Bazaluk, B., Hamdan, M., Ghaleb, M., Gismalla, M.S.M., Correa da Silva, F.S., and Batista, D.M. (2024, January 20–24). Towards a Transformer-Based Pre-trained Model for IoT Traffic Classification. Proceedings of the 2024 IEEE Network Operations and Management Symposium (NOMS), Seoul, Republic of Korea.10.1109/NOMS59830.2024.10575448 ↗
11. Al Hanif, A., and Ilyas, M. (2024). Enhance the Detection of DoS and Brute-Force Attacks within the MQTT Environment through Feature Engineering and Employing an Ensemble Technique. arXiv.10.5121/ijaia.2024.15401 ↗
12. Choi, S., and Cho, J. (2022). Novel Feature-Extraction Method for Detecting Malicious MQTT Traffic Using Seq2Seq. Appl. Sci., 12.10.3390/app122312306 ↗
13. Vaccari, I., Chiola, G., Aiello, M., Mongelli, M., and Cambiaso, E. (2020). MQTTset, a New Dataset for Machine Learning Techniques on MQTT. Sensors, 20.10.3390/s20226578 ↗
14. Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., and Bengio, Y. (2014, January 8–13). Generative Adversarial Nets. Proceedings of the Advances in Neural Information Processing Systems (NeurIPS), Montreal, QC, Canada.
15. Radford, A., Narasimhan, K., Salimans, T., and Sutskever, I. (2025, August 22). Improving Language Understanding by Generative Pre-Training. Available online: https://cdn.openai.com/research-covers/language-unsupervised/language_understanding_paper.pdf.
16. MQTT-based IDS Evasion Method using GPT
17. Mitchell, E., Lee, Y., Khazatsky, A., Manning, C.D., and Finn, C. (2023, January 23–29). DetectGPT: Zero-Shot Machine-Generated Text Detection Using Probability Curvature. Proceedings of the 40th International Conference on Machine Learning (ICML), Honolulu, HI, USA.
18. Tian, E. (2025, August 22). GPTZero. Available online: https://gptzero.me.
19. Elaziz, M.A., Fares, I.A., Dahou, A., and Shrahili, M. (2025). Federated learning framework for IoT intrusion detection using tab transformer and nature-inspired hyperparameter optimization. Front. Big Data, 8.10.3389/fdata.2025.1526480 ↗
20. A Simple Framework to Enhance the Adversarial Robustness of Deep Learning-based Intrusion Detection System10.1016/j.cose.2023.103644 ↗
21. Explainable AI and Random Forest Based Reliable Intrusion Detection System10.1016/j.cose.2025.104542 ↗
22. Vaccari, I., Aiello, M., and Cambiaso, E. (2020). SlowITe, a Novel Denial of Service Attack Affecting MQTT. Sensors, 20.10.3390/s20102932 ↗
23. OpenAI (2025, August 22). GPT-4o: Omni Model for Text, Audio, and Vision. Available online: https://openai.com/index/hello-gpt-4o.