Parallel Micro-Batching and Scalable Inferencing for ML-Based Malicious Traffic Detection

Achmad Basuki, Widhi Yahya, Dzaki R. Malik, Rizal Setya Perdana, Kasyful Amron, Achmad Husni Thamrin, Andrey Ferriyan, Muhammad Niswar

2025

Abstract

Network intrusion detection systems based on machine learning (ML-IDS) face significant challenges in highspeed network environments, such as gigabit-scale packet capture and real-time inference under dynamic traffic conditions. Efficiently handling these challenges is critical to maintaining accurate and timely detection without overwhelming the system resources. This paper presents a scalable ML-IDS architecture featuring a novel parallel micro-batching inference framework integrated with passive optical tapping for non-intrusive traffic monitoring. The proposed inference architecture is critical to achieving a balance between high classification accuracy and computational efficiency. Experimental results demonstrate a 2.65× throughput improvement over traditional sequential processing while maintaining sub-5ms decision times, even under variable traffic loads. Furthermore, the architecture supports horizontal scaling to accommodate growing network demands, ensuring sustained low-latency detection performance. These contributions establish a robust foundation for deploying ML-IDS in high-speed network environments.

Paper Citation

in Harvard Style

Basuki A., Yahya W., Malik D., Perdana R., Amron K., Thamrin A., Ferriyan A. and Niswar M. (2025). Parallel Micro-Batching and Scalable Inferencing for ML-Based Malicious Traffic Detection. In Proceedings of the 1st International Conference on Research and Innovations in Information and Engineering Technology - Volume 1: RITECH; ISBN 978-989-758-784-9, SciTePress, pages 157-163. DOI: 10.5220/0014275200004928

in Bibtex Style

@conference{ritech25,
author={Achmad Basuki and Widhi Yahya and Dzaki R. Malik and Rizal Setya Perdana and Kasyful Amron and Achmad Husni Thamrin and Andrey Ferriyan and Muhammad Niswar},
title={Parallel Micro-Batching and Scalable Inferencing for ML-Based Malicious Traffic Detection},
booktitle={Proceedings of the 1st International Conference on Research and Innovations in Information and Engineering Technology - Volume 1: RITECH},
year={2025},
pages={157-163},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0014275200004928},
isbn={978-989-758-784-9},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Research and Innovations in Information and Engineering Technology - Volume 1: RITECH
TI - Parallel Micro-Batching and Scalable Inferencing for ML-Based Malicious Traffic Detection
SN - 978-989-758-784-9
AU - Basuki A.
AU - Yahya W.
AU - Malik D.
AU - Perdana R.
AU - Amron K.
AU - Thamrin A.
AU - Ferriyan A.
AU - Niswar M.
PY - 2025
SP - 157
EP - 163
DO - 10.5220/0014275200004928
PB - SciTePress