FOAMI: Enhancing ICS Threat Detection via Feature Optimization, Realistic Augmentation, and Mutual Inference
#ICS#IDS#Ensemble#ML
authors Hung T. Nguyen, Kiet T. Le, Nguyen T. Nguyen, Luong N. Ho, Duy B. Vu, Hoa N. Nguyen
venue SOICT 2025 ยท Springer
status Published
pdf https://drive.google.com/file/d/112lOOqM5IZz7-7lsi87NVVfto40NIVKq/view?usp=sharing
Abstract
Proposed FOAMI, a novel intrusion detection framework for Industrial Control Systems (ICS) that combines feature optimization, realistic data augmentation, and a mutual inference ensemble (GBT + DL). Validated on the IEC 60870-5-104 dataset, FOAMI achieved a detection rate of 89.00% and reduced the false alarm rate to 0.99%, outperforming state-of-the-art methods.