Dianlei Xu defends his PhD thesis on Resilient and Sustainable Urban 6G Edge Networks via Learning-based Approaches

M.Eng. Dianlei Xu defends his PhD thesis “Resilient and Sustainable Urban 6G Edge Networks via Learning-based Approaches” on Thursday the 4th of June 2026 at 12 in the University of Helsinki Exactum building, Auditorium CK112 (Pietari Kalmin katu 5, basement). His opponent is Professor Carla Fabiana Chiasserini (Politecnico di Torino, Italy) and custos Professor Pan Hui (University of Helsinki). The defence will be held in English.

The thesis of Dianlei Xu is a part of research done in the Department of Computer Science and in the Systems and Media group at the University of Helsinki. His supervisors have been Professor Pan Hui (University of Helsinki) and Professor Sasu Tarkoma (University of Helsinki).

Resilient and Sustainable Urban 6G Edge Networks via Learning-based Approaches

Urban 6G edge networks are expected to support increasingly diverse and mission-critical services under stringent latency, reliability, and sustainability requirements. The dense integration of communication and computation infrastructure, combined with highly dynamic urban environments, introduces significant uncertainty, large-scale coupling, and complex trade-offs between service resilience and energy efficiency. Designing intelligent control mechanisms that operate reliably at city scale therefore remains a fundamental challenge.

This thesis investigates learning-based approaches for resilient and sustainable control of urban 6G edge networks from a system-level perspective. Rather than treating user-side performance and operator-side efficiency as independent objectives, the thesis argues that resilience and sustainability are interdependent properties that emerge from the dynamic coupling between users, infrastructure, and the surrounding environment. To address this challenge, a unified framework is developed that integrates environment modeling, user-centric decision-making, and infrastructure control within a closed-loop learning paradigm.

At the foundation of the framework lies an AI-driven digital twin that provides a representative and controllable environment for learning-based optimization. Building upon this foundation, the thesis examines risk-aware computation offloading to manage service volatility at the user side, and progressively structured control strategies to improve energy efficiency at the operator side. Across these problem settings, the results demonstrate that learning effectiveness at urban scale depends less on algorithmic complexity than on how decision-making processes are structured to reflect uncertainty, coupling, and system evolution.

By synthesizing these findings, the thesis reframes urban 6G edge networks as systems jointly dominated by risk and coupling, where effective intelligence arises from structural alignment between control architectures and physical interactions. The work contributes a principled perspective on how learning-based network intelligence can be designed to achieve robustness, scalability, and efficiency in complex urban environments, while also identifying the boundaries within which such approaches remain valid. This perspective provides a foundation for future research toward resilient and sustainable networked systems beyond 6G.

Avail­ab­il­ity of the dis­ser­ta­tion

An electronic version of the doctoral dissertation will be available in the University of Helsinki open repository Helda at .

Printed copies will be available on request from Dianlei Xu: .