Agentic AI for Intelligent Digital Twins in Telecom Networks

EPSRC iDLA 4 year Studentship

Supervisors:

• Prof. Ashiq Anjum  aa1180@leicester.ac.uk
• Anthony Conway

Project

The increasing complexity and scale of modern telecom networks, driven by technologies, demand new approaches to monitoring, optimisation, and decision-making. This project explores the integration of agentic artificial intelligence (AI) into digital twin ecosystems to enable future intelligent, semi-autonomous operations within telecom infrastructure, promoting effective human-AI collaboration for safe, explainable decision-making, thus informing the design of the next generation of intelligent, resilient, and sustainable telecom infrastructure.
To investigate how agentic AI can perceive, learn from, and act on the evolving state of telecom digital twins to achieve real-time optimisation, self-healing, and energy-efficient operation, the project will focus on the design of architectures and protocols that enable secure, explainable, and policy-compliant interactions between autonomous agents and distributed network twins that work in coordination with human decision-making.
The project aims to:

• Develop scalable agent-digital twin coordination models for real-time multi-domain network control.

• Enable continual learning within agents while ensuring alignment with telecom policies and service-level agreements.

• Design frameworks for effective human-AI collaboration for safe, explainable decision-making.

• Quantify the energy and sustainability impact of agent-driven operational strategies in dense network environments.

Research Questions
1. How can agentic AI autonomously coordinate with digital twins to optimise real-time network performance across multi-domain telecom environments?

Focus: Real-time decision-making, autonomous agents, orchestration.

Relevance to BT: Telecom networks are multi-layered (core, edge, access) and dynamically evolving. Agentic AI could shift from passive monitoring to proactive and collaborative optimisation.

2. What architectures and communication protocols are needed to ensure robust interaction between digital twins and agentic AI in distributed telecom infrastructures?

Focus: System design, interoperability, latency sensitivity.

Relevance to BT: Telecom systems are geographically distributed and heterogeneous. Defining scalable, interoperable architectures is foundational.

3. How can agentic AI systems in digital twins be designed to support effective human-in-the-loop collaboration, ensuring transparency, trust, and controllability in telecom network operations?

Focus: Human-AI collaboration, trustworthy AI, AI explainability

Relevance to BT:  Intelligent telecom infrastructure will need to balance autonomy with human oversight, ensuring that AI decisions are interpretable, aligned with operator goals, and adaptable to real-world constraints.
Outline the work the student will undertake

• Conduct original research in digital twins, artificial intelligence, human-AI collaboration and distributed systems.

• Produce publications and reports which contribute to BT's world-class standing in telecommunications and ICT research. 

• Contribute to the creation of intellectual property rights, which will enhance BT's competitive position.

The project focuses on developing semi-autonomous agentic AI systems that promote effective human-AI collaboration for safe, explainable decision-making, aligning with EPSRC's goals in trustworthy AI for complex systems like telecom networks. It promotes research in future communication systems, emphasising network automation, resilience, and energy efficiency for enhanced digital connectivity and UK economic growth. Additionally, the project highlights energy-aware, self-optimising networks aimed at creating sustainable digital infrastructure, in line with the UKRI Net Zero strategy. Lastly, it investigates safe, policy-compliant AI, ensuring responsible innovation in safety-critical sectors such as telecommunications.

1. Artificial Intelligence Technologies Theme
https://www.ukri.org/what-we-do/browse-our-areas-of-investment-and-support/artificial-intelligence-technologies/ 
2. Information and Communication Technologies Theme - Networks and Distributed Systems
https://www.ukri.org/what-we-do/browse-our-areas-of-investment-and-support/information-and-communication-technologies-theme/  
3. Artificial Intelligence and Robotics theme - Trustworthy Autonomous Systems (TAS)
https://www.ukri.org/what-we-do/browse-our-areas-of-investment-and-support/artificial-intelligence-and-robotics-theme/ 
4. Human-Computer Interaction (HCI)
https://www.ukri.org/what-we-do/browse-our-areas-of-investment-and-support/human-computer-interaction/ 
5. Energy Networks
https://www.ukri.org/what-we-do/browse-our-areas-of-investment-and-support/energy-networks/

This research supports BT's strategic initiative to develop autonomous networks by integrating agent-based AI with digital twins. This combination enables intelligent, real-time optimisation and fault management, streamlines cross-domain orchestration, and reduces operational complexity. The technology has the potential to improve service assurance for BT's services by providing predictive insights, facilitating rapid fault detection, and ensuring quick recovery to maintain high reliability. By incorporating continual learning and policy-aware agents and incorporating human-in-the-loop mechanisms, BT can scale automation safely and transparently, ensuring effective human-AI collaboration in critical decision-making processes. This approach also helps address operational efficiency by augmenting human-expertise with intelligent systems offering long-term strategic value to position BT at the forefront of intelligent infrastructure in the UK.

This research will benefit BT operations by facilitating more efficient, resilient, and sustainable network operations through AI-driven automation, while ensuring human-in-the-loop oversight and promoting effective human-AI collaboration for safe, explainable decision-making. It will also support the academic community by advancing interdisciplinary research in AI, cyber-physical systems, and network engineering while providing valuable training opportunities for early-career researchers. Ultimately, UK consumers will gain from more reliable connectivity, improved service quality, and a reduced environmental impact as telecom infrastructure becomes smarter and more energy efficient.