The thesis focuses on extending Federated Cooperative Multi-Task Semantic Communication (Federated-CMT-SemCom) for applications such as LEO Constellations. In particular, considering satellite communication scenarios, e.g., channel effects, noisy parameter updates, and real-world dataset use.

Requirements:

• Background in machine learning/ deep learning
  
  
• Experience with PyTorch
  
  
• Interested in semantic communication, multi-task learning, and federated learning
  
  
• Background in space engineering is a plus