Semantic Communications
Qiyang Zhao, Hang Zou, Mehdi Bennis, Merouane Abdelkader Debbah · 2024
Semantic communication transforms the transmitter and receiver pairs into teacher and student agents, which interact with each other through a semantic representation of the underlying information structure. Such representation must satisfy key properties of minimalism (least sufficient bits), generalizability (across different domains) and efficiency (high fidelity generation). Machine learning plays a key role in semantic communication, where generative artificial intelligence (AI) shows strong capabilities in semantic understanding and reasoning. In this chapter, we provide a holistic view on the role of semantic communication as a powerhouse of native AI networks in 6G and beyond, followed by technical insights into the mathematical theories and technologies of semantic communications leveraging machine learning and generative AI.