Reimagining 5G security through cyber defense innovation

A groundbreaking cybersecurity framework designed to enhance the security of 5G wireless communications has been developed by Hadiseh Rezaei, a Ph.D. student at the University of Portsmouth. The innovation, called FedLLMGuard, integrates advanced artificial intelligence technologies to detect and prevent cyber threats more efficiently across 5G networks.

Developed in collaboration with the University of Portsmouth’s Artificial Intelligence and Data Science (PAIDS) Center, the research was published in Computer Networks in 2025. Rezaei’s work addresses the growing vulnerability of 5G systems, which transmit massive amounts of data at high speed and are increasingly used in sectors such as healthcare and finance. FedLLMGuard combines large language models (LLMs)—AI systems capable of understanding context and linguistic patterns—with federated learning, which allows AI models to improve through distributed data sources without compromising user privacy. This dual approach enables real-time threat detection and adaptive protection against evolving cyberattacks.

According to co-author Dr. Rahim Taheri, Senior Lecturer in Computer Science at the University of Portsmouth, most existing 5G Intrusion Detection Systems (IDS) rely on limited numerical data and fail to capture the deeper contextual information that modern AI can process. FedLLMGuard’s architecture overcomes this limitation by dynamically learning from network behavior while maintaining privacy. In performance tests, the system achieved an accuracy rate of 98.64% and detected threats in just 0.0113 seconds, outperforming all existing models. Experts within the PAIDS Center see this as a step toward integrating AI-driven security measures across critical infrastructures. The research underscores a promising future for adaptive cybersecurity frameworks capable of scaling with the demands of intelligent, data-driven networks.