Join our Team

About this opportunity: 
 
Join Ericsson for a master’s thesis investigating how neural-network-based AI receivers can adapt to changing wireless conditions in live radio access networks. You will explore unsupervised and self-supervised online adaptation strategies that enable a physical shared channel (PxSCH) receiver to maintain performance across varying configurations and propagation environments without requiring labeled training data. 

 
What you will do: 
 

You will review research on online learning, self-supervised training, and domain adaptation for AI receivers. Using field-trial data, you will analyze how receiver performance changes across different PxSCH configurations and channel conditions and identify the causes of performance degradation. You will then design, implement, and evaluate one or more online adaptation methods against a static baseline. The evaluation will consider receiver accuracy, link-level performance such as BLER and throughput, adaptation speed, computational cost, and model stability. The thesis will conclude with a presentation of your results and recommendations to the Ericsson research team. 

The skills you bring: 
 

• Background in electrical engineering, computer science, applied mathematics, or similar
• Machine learning and deep learning (ANN, CNNs, training methodologies)
• Wireless communications fundamentals (OFDM, channel estimation, MIMO)
• Programming proficiency in Python (PyTorch, Jax or TensorFlow)
• Background in signal processing is a plus