Artificial intelligence (AI) and 6G are converging to unlock new possibilities for next-generation connectivity. Within the Federated Telecoms Hubs (FTH), CHEDDAR is leading pioneering research that pushes the boundaries of native AI in telecommunication networks and helps shape the UK’s position in global edge intelligence and connectivity innovation.
This article explores how AI and 6G combine, where CHEDDAR’s research fits, and why its contributions are uniquely significant.
Three Ways AI and 6G Combine
To understand CHEDDAR’s role, it’s important to clarify the three commonly accepted ways AI integrates with 6G radio access networking (RAN) [1]:
- AI and RAN: Computing resources within the RAN are used to serve both AI and radio communications separately, potentially saving resources and energy.
AI for RAN: AI is used to improve radio communications, enhancing spectrum efficiency, user experience, and energy optimisation. This is where most of the global academic research is focused, including much of the work from HASC and TITAN.
- AI on RAN: AI is hosted on the RAN to deliver AI services directly to mobile end users, such as enabling personal, configurable, and sovereign ChatGPT models close to the user. This is where significant CHEDDAR’s research is focused.
CHEDDAR’s Unique AI Research
CHEDDAR’s research is driven by major challenges facing humanity and is structured around three core pillars. Within these, there are several key advances in AI that are distinctive to CHEDDAR’s work:
C1. AI for RAN
CHEDDAR is developing large language models (LLMs) designed specifically to understand telecom network engineering. These models aim to automate future research pipelines, or rapidly configure future RAN services [2].
C2. AI on RAN
CHEDDAR is using LLMs in telecoms to empower users to define new services and self-program them into 6G networks. This approach supports flexible emerging business models and diverse use cases, directly enabling AI on RAN capabilities [3].
C3. Trustworthy AI
To build trust in AI systems, CHEDDAR is developing mechanisms that:
- Explain AI actions for mission-critical services and communications [4].
- Verify AI performance in AI for RAN applications to ensure reliability and safety [5].
Together, these innovations demonstrate CHEDDAR’s leadership in advancing intelligent, adaptable, and trustworthy network systems.
Collaborating Across the Federated Telecoms Hubs
CHEDDAR’s work contributes to the broader goals of the Federated Telecoms Hubs (FTH), which unite 27 leading UK universities and research institutions. Through collaboration with hubs such as HASC and TITAN, the programme accelerates innovation and drives the UK’s competitiveness in AI-enabled 6G networks through evidence-based thought leadership in standards [6].
Whether through a single cross-hub article or separate updates from each hub, this campaign highlights the value of collaboration and showcases the UK’s collective strengths in AI for connectivity.
Next Steps
In the next steps, we are particularly interested in tackling 3 main challenges:
- Can RAN unlock new forms of AI architectures or operations
- How can we achieve mobile and efficient AI-as-a-Service in telecoms
- Can we help businesses unlock the full potential of edge and sovereign AI hosted in telecom networks
To learn more about CHEDDAR’s AI research and its role in shaping the future of intelligent connectivity, visit the CHEDDAR website.
[1] https://www.nvidia.com/en-gb/glossary/ai-ran/
[2] “RFSensingGPT: A Multi-Modal RAG-Enhanced Framework for Integrated Sensing and Communications Intelligence in 6G Networks,” in IEEE Transactions on Cognitive Communications and Networking, 2025
[3] “Automatic Retrieval-Augmented Generation of 6G Network Specifications for Use Cases,” IEEE Communications Magazine, 2024
[4] “Explainable Reinforcement and Causal Learning for Improving Trust to 6G Stakeholders,” IEEE Open Journal of the Communications Society, 2025
[5] “Towards Achieving Energy Efficiency and Service Availability in 6G O-RAN via Formal Verification,” ACM International Symposium, DataMod 2024
[6] “Towards Standardization of GenAI-Driven Agentic Architectures for Radio Access Networks,” Frontiers in Artificial Intelligence, 2025
