Towards Personalised Healthcare: Agentic AI in Federated 6G Wireless Edge Networks

About the Project

The evolution of wireless networks towards 6G is enabling new paradigms for intelligent, low-latency, and privacy-aware applications. In the healthcare sector, this shift supports the transition towards personalised, real-time medical services powered by ubiquitous sensing and intelligent data processing. However, existing cloud-centric models face limitations in latency, bandwidth, and trust, particularly for mission-critical healthcare scenarios.

This PhD project investigates the integration of agentic AI and Mobile Edge Computing (MEC) within 6G-enabled wireless networks to deliver decentralised, autonomous intelligence for personalised healthcare. The core idea is to deploy goal-directed AI agents across edge nodes, such as smart devices and local servers, enabling them to make local decisions, coordinate learning, and adapt in real time to patient and system-level dynamics. These agents will collaborate to provide timely, privacy-preserving insights while reducing reliance on centralised infrastructure. The outcome will be a resilient, scalable framework for autonomous health monitoring and intervention across next-generation wireless networks.

Research Objectives:

• Develop agentic AI models capable of autonomous health data processing, decision-making, and adaptation within the constraints of MEC nodes in wireless network environments.
• Design distributed coordination protocols that enable intelligent agents to collaborate across the edge-cloud continuum, optimising latency, energy efficiency, and data privacy.
• Implement federated or swarm learning techniques suitable for deployment in bandwidth-constrained and privacy-sensitive wireless healthcare networks.
• Evaluate system performance and robustness in real-world healthcare use cases, focusing on responsiveness, trustworthiness, and scalability across diverse network conditions and patient contexts.

Funding Notes

there is no funding for this project