A framework for building tighter security into 5G wireless communications has been created by a Ph.D. student working with the University of Portsmouth’s Artificial Intelligence and Data Center.
With its greater network capacity and ability to rapidly transmit huge amounts of information from one device to another, 5G is a critical component of intelligent systems and services—including those for health care and financial services.
However, the dynamic nature of 5G networks, the high volumes of data shared and the ever changing types of information transmitted means that these networks are extremely vulnerable to cyber threats and increasing risks of attack.
Hadiseh Rezaei, a Ph.D. student at the University of Portsmouth with a background in computer network and software engineering, has considered this issue and carried out experimental research resulting in the development of a framework that could lead to better safeguards around information shared between devices using 5G networks.
This research, published in Computer Networks, proposes a new framework, named FedLLMGuard, which combines two technologies: large language models which understand language and patterns; and federated learning, which is a system that learns from many different sources without anyone having to share private information.
Together, they create a single system which accurately and rapidly detects abnormalities in 5G networks and securely protects data privacy in real-time.
Co-author Rahim Taheri, Senior Lecturer in Computer Science at the University of Portsmouth’s School of Computing, explained, “The majority of 5G Intrusion Detection Systems still rely heavily on the numerical features in TV data, restricting their ability to capture wording, logic or the contextual nuances.
“Large Language Models are a bit like the building blocks of data reading. They are trained on immense amounts of data, making them capable of understanding language and context, but they are still very underused in network security. Whereas, federated learning is a way to train AI models without humans viewing private data, offering a means to unlock information which can be fed into new AI applications.”
Hadiseh added, “Conventional Intrusion Detection Systems often rely on fixed rules or static machine learning models. Separately, these approaches struggle to handle the constantly changing nature of 5G traffic and are not effective against new or sophisticated attacks. However, FedLLMGuard dynamically adapts to protect against new threats as they emerge.
“Through our experiments, we have demonstrated that by bringing federated learning together with large language models, 5G security can be accurately improved, at speed. Think of it as being like a super-smart security guard for the internet that never gets tired, learns from every new trick hackers try, and protects everyone’s private information at the same time.”
To prove how robust and reliable FedLLMGuard is, researchers tested it against various cybersecurity threats. The framework successfully defended against conflicting manipulation attempts, large-scale cyber attacks, stealth attacks designed to slip past security systems undetected, and data poisoning attacks that try to corrupt the AI training process.
FedLLMGuard outperformed all models, achieving accuracy of 98.64% in recognizing a threat to security, at a speed of under 0.02 of a second ( 0.0113s). The results demonstrated FedLLMGuard’s capacity to detect threats rapidly, mitigate attacks effectively, and sustain high accuracy while ensuring data privacy, make it a scalable and resource-efficient security solution for 5G networks.
Recognizing that artificial intelligence (AI) and data science are rapidly advancing fields within research and innovation, the University of Portsmouth formally launched the Portsmouth AI and Data Science (PAIDS) Center in June 2025.
At the core of the PAIDS Center is the development of computer patterns, methods, and algorithms to create solutions for improving systems and delivery of services in areas including health and well-being, education, cybersecurity and digital marketing.
More information:
Hadiseh Rezaei et al, FedLLMGuard: A federated large language model for anomaly detection in 5G networks, Computer Networks (2025). DOI: 10.1016/j.comnet.2025.111473
University of Portsmouth
Citation:
Cyber defense innovation could significantly boost 5G network security (2025, October 8)
retrieved 8 October 2025
from https://techxplore.com/news/2025-10-cyber-defense-significantly-boost-5g.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.
