CIRCAT at IEEE CloudCom 2026 workshops

Prevention is cheaper than a breach

CIRCAT partner K3Y LTD (K3Y) will present two research papers at IEEE CloudCom 2026 workshops, taking place in Paris, France, on 27–29 October 2026.  

The first paper, “Evaluating Federated Learning Tree Ensembles for 5G Intrusion Detection under Extreme Data Heterogeneity,” will be presented on Tuesday, 27 October 2026, during Session 1: Cloud, Network & Fog Security, at the SCC 2026 – International Workshop on Secure Cloud Continuum.  

The paper examines whether tree-based models such as XGBoost, which perform best in centralized intrusion detection, keep their performance in federated 5G edge settings. It shows that they lose more than 70% of their performance when edge nodes observe different types of attacks, while neural networks and linear models remain considerably more stable, providing guidance on which models suit decentralised security deployments. 

The second paper, “Tackling Data Heterogeneity at the 5G Edge: An Architectural Co-Design for Federated Intrusion Detection Systems,” will be presented on Wednesday, 28 October 2026, during Session 1: Federated Intrusion Detection, Digital Twins & Trustworthy Architectures, at the CRSN 2026 workshop – Building a Resilient Europe: Integrating AI and Cybersecurity across Critical Infrastructures.  

The paper studies how AI-based intrusion detection can be trained collaboratively across 5G edge nodes with Federated Learning, without sharing raw network traffic. By benchmarking ten federated strategies across three model families on real 5G traffic, it shows that the model architecture and the aggregation strategy must be chosen together, and that simpler neural networks are more robust when edge nodes observe very different traffic. The work concludes with practical deployment guidelines for 5G edge security. 

Both papers acknowledge CIRCAT funding and contribute to the project’s work on AI-based threat detection for critical infrastructures, with a focus on privacy-preserving and collaborative intrusion detection in 5G networks. 

 

📩 Stay updated with CIRCAT → Subscribe to our newsletter here.

👉Follow us on LinkedIn and Facebook

Scroll to top

Stay updated with our project

Get the latest updates, and events from our project - directly in your inbox.