Digital Twin-in-the-Loop Resource Optimization towards AI-Native 6G Radio Access Networks

Description

Next-generation wireless networks, such as 5G-Advanced and 6G, must support increasingly demanding use cases, including immersive communications, integrated sensing and localization, autonomous mobility, industrial automation, and digital health services. Meeting these diverse requirements calls for a fundamental transformation in how networks are operated and managed. Future wireless systems will need to incorporate multimodal contextual awareness, real-time situational intelligence, closed-loop control, and rapid “what-if” analysis to ensure proactive and optimal performance in highly dynamic environments.

Digital Twin Networks (DTNs)—defined as virtual replicas of physical radio infrastructures—represent a promising approach to achieving this advanced level of intelligent network management. Unlike traditional simulation tools that rely on static statistical models, a DTN continuously mirrors the real network by accurately modeling RF propagation in realistic 3D environments, integrating full protocol-stack behavior, and maintaining bidirectional synchronization with the physical system. These capabilities make DTNs powerful tools for site-specific AI training, proactive performance monitoring, interference management, and policy evaluation.

However, building and maintaining a high-fidelity DTN remains challenging due to the scale, dynamics, and heterogeneity of modern networks. Existing platforms, such as Microsoft Azure Digital Twins, focus primarily on data collection and visualization, with limited capabilities for prediction or closed-loop control. Conversely, emerging tools like NVIDIA Sionna provide realistic channel modeling and link-level performance evaluation, but lack seamless integration with real-time feedback mechanisms.

This internship project aims to bridge this gap by developing a closed-loop Network Digital Twin (NDT-in-the-loop) using the software tools and testbed available at the Radiocommunications and Signal Processing Laboratory (LRTS) of the Department of Electrical and Computer Engineering at Université Laval.

The project tasks include:
- designing an NDT-in-the-loop system using Azure Digital Twin and a real-time simulator based on Sionna;
- developing a generative AI and continual learning strategy to maintain synchronization while reducing communication overhead;
- validating the NDT-in-the-loop system on a hardware testbed.

The expected outcome is a scalable DTN-assisted control platform capable of providing real-time situational awareness, robust radio-resource optimization, and resilience to unforeseen conditions—an essential building block for intelligent automation in 6G networks.

Requirements and Conditions

The candidate needs to be present in Canada.