TelCoSec4Lux

5G networks are reshaping telecommunications with unprecedented connectivity and speed, but this same openness exposes operators to growing security risks: data breaches, unauthorized access, manipulated content, and increasingly sophisticated intrusion attempts that traditional tools struggle to catch in real time. Building on the 5G Secure Experience (Secure5GExp) project (2020–2023), which delivered a Telecom Intrusion Detection System (TIDS) operating within live 5G networks, this new initiative tackles the next challenge: securing the entire lifecycle of telecom data, from generation to storage.

TelCoSec4Lux aims to develop an end-to-end security framework combining stronger data privacy, dynamic threat detection, advanced access control, and real-time fake-content detection in multimedia data — equipping telecom operators to move from passive awareness to proactive, strategic protection of their networks.

This project is a research-driven initiative that builds an end-to-end security framework covering the full lifecycle of telecom data, extending the results of the 5G Secure Experience project. The project will:

• Design privacy-preserving, secure federated machine learning solutions for telecom data.

• Develop cross-protocol ML capabilities to detect and mitigate evolving 5G threats.

• Build deep-learning models to detect fake content (voice, photos, videos) in multimedia data.

• Develop an AI assistant, powered by a fine-tuned LLM, to help telecom operators understand and act on security risks.

• Deliver Proof-of-Concepts (PoCs) demonstrating privacy-preserving detection, fake-content identification, and AI-driven risk reporting to real customers.

This multidisciplinary approach, combining federated learning, threat intelligence, fake-content detection, and human-centric AI assistance, sets the project apart from traditional, reactive telecom security tools.

LIST leads the research and development of privacy-preserving federated ML solutions and cross-protocol threat detection models, and builds the fine-tuned LLM-based AI assistant demonstrated to customers.

The project's outcomes will provide:

• A privacy-preserving federated machine learning framework for secure telecom data processing.

• Cross-protocol threat detection models, strengthening resilience against evolving 5G-era attacks.

• A deployable fake-content detection prototype, integrated into LuxID, to identify manipulated voice, photo, and video content.

• A functional AI assistant, demonstrated to real customers, turning technical risk data into actionable insights.

By strengthening data privacy, threat detection, and content authenticity, the project will help telecom operators shift from reactive defense to proactive protection — with real-world applications spanning secure infrastructure, fraud and disinformation detection, and customer data protection.