Research
Extreme Edge Computing
Pushing the boundaries of theory and design to drive tomorrow’s innovations.
Queen’s TRL seeks to propel the boundaries of computation beyond the conventional solutions of cloud and edge servers. Our research in Extreme Edge Computing (XEC) is centered around harnessing the immense, often untapped, resources of devices at the network's very edge.
In close proximity to the end-user, smartphones, vehicles, and other IoT devices can be harnessed to create a decentralized and democratized computing fabric that delivers superior performance, enhanced user privacy, whilst enabling powerful on -and near-device applications - all without burdening the backhaul network.
To achieve this, we design and analyze foundational XEC architectures, creating intelligent and proactive algorithms to orchestrate these heterogeneous devices. Such orchestration could involve resource-aware scheduling, offloading, and replication. By modelling system dynamics and user behaviour, our solutions anticipate service demands and optimally place computational tasks, ensuring seamless service continuity and Quality of Service.
Our research fuses rigorous theoretical modelling with practical, hands-on implementation. Quantifying performance trade-offs and augmenting real-world distributed systems using reactive and proactive techniques allows us to overcome the challenges of resource-constrained XEC environments.
Intelligent Management of Next-Generation Wireless Networks
Relevance through Industry partnership, Leadership through discourse and engagement.
Next-generation wireless networks (NGWNs) require a shift from rule-based management to more complex and intelligent, data-driven automation. The radio access network (RAN) provides the access layer connecting user devices to NGWNs. It is often more challenging than other domains due to hard real-time deadlines, non-stationarity, and partial observability of the wireless environment, as well as local actions with network-wide ripple effects. At Queen’s TRL, we see the RAN as the invisible fabric of our connected world, linking people and devices in a web shaped by artificial intelligence (AI) technologies. As RAN technology advances toward 6G, unexplored frontiers await AI-driven innovation to enhance performance and optimize resource utilization.
Queen’s TRL designed and tailored machine learning (ML) techniques create autonomous networks that are both efficient and resilient, ensuring that 6G systems can deliver on their promises of performance and reliability. Key research activities focus on trustworthy AI for radio resource management (RRM) functions, such as network slicing and handover management. We develop agile AI frameworks, able to meet the challenges of NGWNs through strong generalization, robustness, safety, and explainability. We also explore versatile wireless foundation models: pretrained representations learned from network measurements, logs, and traces. Rather than being defined by size, these models are built for transfer and can be adapted to new cells, bands, and vendors, enabling zero- or few-shot generalization. This supports rapid improvements in resource optimization without per-site retraining.
At Queen’s TRL, partnering with Industry is central to our work, demonstrated through collaborations with partners such as Ericsson Canada, and engagement with open standards such as the O-RAN architecture. Queen’s TRL continues to lead through the delivery of tutorials at prominent IEEE conferences, engaging and shaping the global discourse on the future of AI in networking.
