Machine Learning-based Modulation and Coding Design

This talk explores the transformative potential of machine learning in redesigning fundamental physical layer components - modulation and coding schemes - for next-generation wireless systems. We present novel data-driven approaches that challenge traditional model-based designs, demonstrating how neural networks can autonomously learn optimal modulation constellations and error correction coding strategies tailored to dynamic channel conditions. The discussion covers practical implementation challenges, including computational complexity, training data requirements, and hardware compatibility.