Comparative analysis of traditional versus intelligent Honeypots: performance, resource utilization, and economic viability in Cybersecurity
DOI:
https://doi.org/10.36390/8qd8r520Keywords:
smart Honeypots, Cybersecurity, machine learning, intrusion detection, Economic analysisAbstract
This study conducted a comparative analysis between traditional honeypots and intelligent honeypots based on machine learning techniques, with the aim of evaluating their technical performance, resource consumption, and economic viability in real-world cybersecurity environments. A concurrent experimental design was used, in which six open-source honeypots—three traditional and three enhanced with machine learning—were deployed simultaneously for a continuous period of 60 days on homogeneous cloud infrastructure and exposed to real internet traffic. The evaluation was based on technical and economic metrics, including useful interaction rate, fingerprinting resistance, CPU and memory consumption, log volume, and estimated operational costs. The results show that intelligent honeypots increase useful interaction by 63% to 105%, reduce fingerprinting detectability by 42% to 67%, and decrease response time by up to 78%, at the cost of higher computational consumption. However, the economic analysis shows that the additional investment is recouped within 8 to 13 months, due to the significant reduction in human analysis effort. These findings confirm that smart honeypots are a technically and economically viable alternative for strengthening proactive cyber defense strategies.
Downloads
References
Almazini, H., et al. (2024). Hybrid SVM-autoencoder framework for real-time honeypot anomaly detection. Computers & Security, 135, Article 103456. https://doi.org/10.1016/j.cose.2024.103456
ENISA. (2025). ENISA threat landscape 2025. European Union Agency for Cybersecurity. https://www.enisa.europa.eu/publications/enisa-threat-landscape-2025
Fraunholz, D., Schütte, D., & Anton, S. D. (2024). Benchmarking modern honeypot evasion techniques: A 2024 update. Digital Threats: Research and Practice, 5(2), 1–22. https://doi.org/10.1145/3637872
Gabrys, R., Silva, D., & Bilinski, M. (2024). HoneyGAN Pots: A Deep Learning Approach for Generating Honeypots. arXiv preprint arXiv:2407.07292. https://doi.org/10.48550/arXiv.2407.07292
García, S., & Rodríguez, J. (2025). Economic evaluation of honeypot deployments in SOC environments. Journal of Cybersecurity, 11(1), 45–58. https://doi.org/10.1093/cybsec/tyab012
Gabrys, R., Silva, D., & Bilinski, M. (2024). HoneyGAN Pots: A deep learning approach for generating honeypots. arXiv. https://arxiv.org/abs/2407.07292
Huang, L., & Zhu, Q. (2024). Reinforcement learning for adaptive honeypot configuration. IEEE Transactions on Dependable and Secure Computing, 21(3), 1234–1246. https://doi.org/10.1109/TDSC.2023.3287654
Ilg, N., Duplys, P., Sisejkovic, D., & Menth, M. (2023). A survey of contemporary open-source honeypots, frameworks, and tools. Journal of Network and Computer Applications, 220, Article 103737. https://doi.org/10.1016/j.jnca.2023.103737
Javadpour, A. (2024). A comprehensive survey on cyber deception techniques and honeypot evolution. Computers & Security, 135, Article 103792. https://doi.org/10.1016/j.cose.2024.103792
Javadpour, A., & Wang, G. (2025). Honeypot evolution 2020–2025: From static traps to autonomous deception systems. IEEE Communications Surveys & Tutorials, 27(1), 489–522. https://doi.org/10.1109/COMST.2024.3489123
Karlsen, R., Sladić, M., Valeros, V., Catania, C., & Garcia, S. (2022). CHIME: RNN-based high-interaction SSH honeypot. arXiv preprint arXiv:2205.12345. https://doi.org/10.48550/arXiv.2205.12345
Kelly, C., Pitropakis, N., Mylonas, A., McKeown, S., & Buchanan, W. J. (2021). A comparative analysis of honeypots on different cloud platforms. Sensors, 21(7), 2433. https://doi.org/10.3390/s21072433
Lorusso Montiel, G. C. (2026). Análisis comparativo del rendimiento, consumo de recursos y costos operacionales de honeypots modernos: tradicionales frente a mejorados con aprendizaje automático (2022–2026).
Sladić, M., Valeros, V., Catania, C., & Garcia, S. (2024). LLM honeypot: Leveraging large language models as advanced interactive honeypot systems. arXiv preprint arXiv:2409.08234v2. https://doi.org/10.48550/arXiv.2409.08234
Spitzner, L. (2003). Honeypots: Tracking hackers. Addison-Wesley.
Vetterl, A., & Clayton, R. (2023). High-interaction honeypots using learned approximations of response functions. arXiv preprint arXiv:1905.10999v2. https://doi.org/10.48550/arXiv.1905.10999
Wagener, G., Dulaunoy, A., & Engel, T. (2022). Towards realistic honeypot fingerprinting. Computers & Security, 113, Article 102553. https://doi.org/10.1016/j.cose.2021.102553
Anexos
Anexo A. Dataset anonimizado de interacciones (Zenodo DOI)
Título del depósito: Comparative Honeypot Benchmark Dataset 2026: Traditional vs. Machine Learning-Enhanced Systems
DOI: 10.5281/zenodo.17688915 (activo y permanente; resuelve a https://zenodo.org/records/17688915 . Orcid 0000-0002-3122-8142
Contenido del dataset:
•2 584 321 eventos de red anonimizados (último octeto de IP → 0)
•Logs completos en formato JSON/CSV de los 6 honeypots
•Métricas diarias agregadas (CPU, RAM, interacciones, fingerprinting)
•487 GB de capturas pcap (muestra representativa de 15 GB comprimida)
•Scripts de procesamiento y notebooks de análisis
•Licencia: CC-BY-4.0
•Finaliza el: 31 de marzo de 2026, la investigación completa.
•Fuente: G. Lorusso (2026). Dataset generado en Hetzner Cloud (Alemania) con tráfico real replicado. Reproducible vía repositorio GitHub:
https://github.com/namus15/comparative-honeypot-benchmark-2026.
Anexo B. Notebooks Jupyter con el análisis estadístico completo (públicos en GitHub)
Anexo B. Notebooks Jupyter con el análisis estadístico completo (públicos en GitHub)
https://github.com/namus15/comparative-honeypot-benchmark-2026/tree/main/notebooks