Master Thesis: Design and Optimization of a Dynamic Reduced-Order Model for Performance - Radio Unit

Join our Team

About this opportunity

Ericsson is offering a Master thesis opportunity within Thermal System Performance in Lund, Sweden. The work focuses on developing and validating a computationally efficient thermal digital twin for fast prediction of transient temperature behaviour in telecom radio units under realistic traffic scenarios.

As mobile communication traffic becomes increasingly dynamic and energy-saving features are introduced, understanding temperature evolution over time is becoming more important. High-fidelity commercial electronics-cooling simulations can describe this behaviour, but they often require significant computational effort and are not always practical for repeated or near real-time analysis.

In this thesis, you will develop a dynamic reduced-order thermal model that can be implemented in MATLAB, Simulink, Python, or an equivalent environment for rapid analysis of the time-dependent thermal behaviour of a telecom radio unit.

You will use available simulation models and commercial simulation software to generate input and reference data, then calibrate and validate the reduced-order model against high-fidelity simulation results.

The work includes investigating the trade-off between model fidelity and computational performance, with the aim of enabling fast or near real-time thermal prediction of complex telecom radio units.

Possible modelling approaches include RC-network representations, state-space formulations, system-identification methods, Proper Orthogonal Decomposition, or hybrid physics-informed data-driven methods. You will select and justify the final approach based on accuracy, interpretability, implementation complexity, and suitability for dynamic thermal prediction.

You will also assess how modelling assumptions, model structure, parameter selection, and numerical stability influence the ability to reproduce transient thermal behaviour, including the balance between computational speed, prediction accuracy, and physical relevance.

What you will do

• Perform a literature study covering transient thermal modelling, reduced-order modelling, and relevant digital-twin approaches.

• Develop, implement, calibrate, and validate a dynamic reduced-order thermal model.

• Compare the model with high-fidelity simulation data using relevant accuracy, stability, and performance metrics.

• Analyse sensitivity, robustness, assumptions, limitations, and validity range under different operating conditions.

• Document the results and provide recommendations for how the model can be further developed or applied.

The skills you bring

You are a Master’s student in Mechanical Engineering, Engineering Physics, Applied Physics, or a related field. Knowledge of heat transfer, fluid mechanics, numerical methods, system modelling, or control theory is beneficial. Experience with MATLAB, Simulink, Python, or equivalent tools is an advantage, together with a strong interest in thermal modelling, simulation, and applied product development.