Adaptive Defense Mechanisms: A Federated Learning Approach for Proactive Intrusion Detection in Heterogeneous IoT Networks

Zainal Syahlan (Sekolah Tinggi Teknologi Angkatan Laut)
Ethan Tan (National University of Singapore)

Journal of Social Science Utilizing Technology
Vol. 4 No. 2 (2026)

Background. The rise of heterogeneous IoT devices has increased security risks, but traditional intrusion detection systems struggle with the diversity and limited resources of these devices. Purpose. This research investigates Federated Learning (FL) to develop a decentralized, adaptive IDS that enables collaborative threat detection while ensuring data privacy and low computational load. Method. An FL model was implemented in a simulated IoT network featuring sensors and industrial controllers, then tested against DoS and data injection attacks using accuracy and resource metrics. Results. The FL-based IDS reached a detection accuracy of 95.3% with minimal resource consumption, proving its efficiency for resource-constrained IoT environments. Conclusion. Federated Learning provides a scalable and proactive solution for IoT security, offering a robust framework for privacy-preserving and efficient intrusion detection.