My paper “Federated Learning for Malware Detection in IoT Devices” has been honored with the 2024 Best Paper Award by the journal Computer Networks. This recognition underscores the significance of our work in advancing cybersecurity methodologies for the Internet of Things (IoT).

In this research, we explore the application of federated learning for detecting malware in IoT devices. Traditional centralized approaches often face challenges related to privacy concerns and the heterogeneous nature of IoT environments. Our study proposes a decentralized method that enables IoT devices to collaboratively learn a shared detection model while keeping the data localized, thus preserving privacy and reducing communication overhead.

We evaluated our framework using the N-BaIoT dataset, which models network traffic of several real IoT devices under malware attacks. Both supervised and unsupervised federated models were trained and assessed, demonstrating that federated approaches can achieve performance comparable to centralized methods while maintaining data privacy.

Furthermore, we investigated the robustness of federated learning against adversarial attacks, specifically data poisoning scenarios. Our findings highlight the vulnerabilities of standard aggregation methods and propose alternative aggregation functions that enhance resilience against malicious participants.

Authors

• Valerian Rey
• Pedro Miguel Sánchez Sánchez
• Alberto Huertas Celdrán
• Gérôme Bovet

Further Information

🏆 Official announcement on Elsevier’s website

📄 Direct link to the article

📰 News post by CyberDataLab