Background and Motivation
As mobile network demand continues to increase due to higher data consumption and densification of user equipment, traditional cellular infrastructures struggle to keep up with both performance and sustainability expectations. Today's networks are often managed through static configurations, such as fixed transmission power, handover thresholds, and scheduling parameters, that are manually tuned and seldom adapt to fluctuating traffic patterns or environmental changes. This results in over-provisioned, energy-inefficient deployments that cannot dynamically respond to varying loads or optimize resource allocation in real time.
To address these limitations, telecom operators like VodafoneZiggo are exploring ways to evolve from static to adaptive network configurations and deployments. The project aims to understand how configurations and designs can be made responsive to real-world network dynamics, thereby improving energy efficiency and user experience. In particular, this thesis focuses on identifying which existing static configurations in VodafoneZiggo's operational network have the most potential for dynamic adjustment, and how machine learning (e.g., online learning, reinforcement learning) or optimization techniques could facilitate such transitions.
The unique setting of this research, with the student embedded in VodafoneZiggo's team, offers direct access to practical data, domain expertise, and real-world challenges that go beyond academic simulation. This tight integration between academic inquiry and industrial application is critical to ensuring future networks are not just faster, but also greener and smarter.
Expected Outcomes
The thesis will begin with a deep-dive analysis of VodafoneZiggo's current cellular network configurations to understand which parameters (e.g., cell switch-off policies, power scaling, load balancing rules) are currently static and how they impact network responsiveness and energy usage.
The student will work to:
• Map and categorize key static parameters and design choices currently in use in the radio access network (RAN) and core network configurations;
• Analyze historical and real-time network traffic and performance data to identify inefficiencies or slow adaptations;
• Explore candidate dynamic strategies, such as online learning, threshold optimization, or multi-objective control frameworks, to improve reactivity and energy usage;
• Design a conceptual framework or prototype simulation that shows how such dynamic configurations could be integrated and evaluated (e.g., in a testbed or emulated environment);
• Develop evaluation metrics such as reconfiguration speed, energy savings, and quality-of-service preservation under dynamic conditions.
Requirements
• Proficiency in Python (for data analysis, modeling, or simulation).
• Familiarity with networking fundamentals, particularly in cellular systems- knowledge of 5G architectures is a plus;
• Understanding of Linux environments - knowledge of containerization (e.g., Docker) is a plus;
• Background in machine learning and data analysis;
• Interest in optimization, machine learning (especially online), or systems modeling.
• Willingness to work with real-world datasets and possibly interact with operational tooling within VodafoneZiggo.
Related Work
Michail Kalntis, José Suárez-Varela, Jesús Omaña Iglesias, Anup Kiran Bhattacharjee, George Iosifidis, Fernando A. Kuipers, and Andra Lutu. 2024. Through the Telco Lens: A Countrywide Empirical Study of Cellular Handovers. In Proceedings of the 2024 ACM on Internet Measurement Conference (IMC '24). Association for Computing Machinery, New York, NY, USA, 51–67. https://doi.org/10.1145/3646547.3688452
Michail Kalntis, Andra Lutu, Jesús Omaña Iglesias, Fernando A. Kuipers, and George Iosifidis. 2025. Smooth Handovers via Smoothed Online Learning. In Proceedings of the 2025 ACM on Internet Measurement Conference (IMC '25). Association for Computing Machinery, New York, NY, USA. https://arxiv.org/abs/2501.08099
Viktoria‑Maria Alevizaki, Markos Anastasopoulos, Anna Tzanakaki, and Dimitra Simeonidou. 2021. Dynamic Selection of User Plane Function in 5G Environments. In Proceedings of the 2021 International Conference on Optical Network Design and Modeling (ONDM '21). IFIP, Athens, Greece. https://dl.ifip.org/db/conf/ondm/ondm2021/1570718826.pdf
Irian Leyva‑Pupo, Cristina Cervelló‑Pastor, Christos Anagnostopoulos, and Dimitrios P. Pezaros. 2022. Dynamic UPF Placement and Chaining Reconfiguration in 5G Networks. Computer Networks, vol. 215. Elsevier. DOI:10.1016/j.comnet.2022.109200 https://www.sciencedirect.com/science/article/pii/S1389128622002900
Endri Goshi, Hasanin Harkous, Shohreh Ahvar, Rastin Pries, Fidan Mehmeti, and Wolfgang Kellerer. 2024. Joint UPF and Edge Applications Placement and Routing in 5G & Beyond. IEEE MASS Workshop on Mobile Access, Sharing & Services (MASS '24). IEEE. https://arxiv.org/abs/2407.07669
Ramy Mohamed, Marios Avgeris, Aris Leivadeas, and Ioannis Lambadaris. 2024. Online VNF Placement using Deep Reinforcement Learning and Reward Constrained Policy Optimization. In Proceedings of the 2024 IEEE International Mediterranean Conference on Communications and Networking (MeditCom '24). IEEE, 1–6. DOI: 10.1109/MeditCom61057.2024.10621193
