Quantum-Resilient Federated Learning for Multi-Layer Cyber Anomaly Detection in UAV Systems
Sensors · 12.01.2026 · 26 (2) · 509
DOI: 10.3390/s26020509 ↗Kurz gefasst
Fachartikel in der Zeitschrift Sensors, online erschienen am 12.01.2026. Verfasst von Canan Batur Şahin. Beteiligte Einrichtung: Faculty of Engineering and Natural Sciences, Malatya Turgut Özal University, Malatya 44900, Turkey. Die Arbeit wurde bisher 6-mal zitiert (Crossref) und nennt 33 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Unmanned Aerial Vehicles (UAVs) are increasingly used in civilian and military applications, making their communication and control systems targets for cyber attacks. The emerging threat of quantum computing amplifies these risks. Quantum computers could break the classical cryptographic schemes used in current UAV networks. This situation underscores the need for quantum-resilient, privacy-preserving security frameworks. This paper proposes a quantum-resilient federated learning framework for multi-layer cyber anomaly detection in UAV systems. The framework combines a hybrid deep learning architecture. A Variational Autoencoder (VAE) performs unsupervised anomaly detection. A neural network classifier enables multi-class attack categorization. To protect sensitive UAV data, model training is conducted using federated learning with differential privacy. Robustness against malicious participants is ensured through Byzantine-robust aggregation. Additionally, CRYSTALS-Dilithium post-quantum digital signatures are employed to authenticate model updates and provide long-term cryptographic security. Researchers evaluated the proposed framework on a real UAV attack dataset containing GPS spoofing, GPS jamming, denial-of-service, and simulated attack scenarios. Experimental results show the system achieves 98.67% detection accuracy with only 6.8% computational overhead compared to classical cryptographic approaches, while maintaining high robustness under Byzantine attacks. The main contributions of this study are: (1) a hybrid VAE–classifier architecture enabling both zero-day anomaly detection and precise attack classification, (2) the integration of Byzantine-robust and privacy-preserving federated learning for UAV security, and (3) a practical post-quantum security design validated on real UAV communication data.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Faculty of Engineering and Natural Sciences, Malatya Turgut Özal University, Malatya 44900, Turkey.
ORCID-Profile: Canan Batur Şahin
Zitieren
APA
Şahin, C. B. (2026). Quantum-Resilient Federated Learning for Multi-Layer Cyber Anomaly Detection in UAV Systems. Sensors, 26(2), 509. https://doi.org/10.3390/s26020509
Vancouver
Şahin CB. Quantum-Resilient Federated Learning for Multi-Layer Cyber Anomaly Detection in UAV Systems. Sensors. 2026;26(2):509. doi:10.3390/s26020509
BibTeX
@article{ahin2026,
author = {Şahin, Canan Batur},
title = {Quantum-Resilient Federated Learning for Multi-Layer Cyber Anomaly Detection in UAV Systems},
journal = {Sensors},
year = {2026},
volume = {26},
number = {2},
pages = {509},
doi = {10.3390/s26020509}
}
EUVIMED