One Computer Science MSc thesis to be presented on 28 May

Wednesday, 28 May there will be a master thesis presentation in Computer Science at Lund University, Faculty of Engineering.

The presentation will take place in E:4130 (Lucas).

Note to potential opponents: Register as an opponent to the presentation of your choice by sending an email to the examiner for that presentation (firstname.lastname@cs.lth.se). Do not forget to specify the presentation you register for! Note that the number of opponents may be limited (often to two), so you might be forced to choose another presentation if you register too late. Registrations are individual, just as the oppositions are! More instructions are found on this page.

14:15-15:00 in E:4130 (Lucas)

Presenters: Hampus Lindqvist, Ellen Petersen
Title: Modeling 5G Network Energy Consumption Caused by Connected Devices
Examiner: Mathias Haage
Supervisors: Maj Stenmark (LTH), Niclas Palm (Ericsson), Ali Nader (Ericsson)

As the energy consumption has increased significantly with 5G compared to its predecessors, the need for optimizations and demand for detailed reporting has grown. Alongside this, network operators are requesting emission reports on a per-user level instead of the network's total energy usage. Due to the lack of energy meters that can measure this, there is a need for a model that can distribute the total energy consumption among the connected users. To address this challenge, we developed and evaluated a machine learning-based approach to model network energy consumption caused by connected user equipment (UE) in 5G base stations. The evaluation consists of a comparison between three different machine learning models and one mathematical model. The thesis included three main parts: the machine learning models' ability to predict the total energy consumption, the distribution of energy usage among connected UEs with machine learning-based models, and an analysis of how specific UE behaviors affect the network energy usage.

Our findings show that XGBoost outperforms both the Transformer and ANN with a R2-score of above 0.97 and Mean Squared Error below 7.5. Together with a proportional distribution formula, it also shows the most promise as a UE attribution model, especially given future advancements of 5G base stations.

Link to popular science summary: To be uploaded