• DOI:10.48550/arXiv.2211.14343
• Corpus ID: 254043722

Less Data, More Knowledge: Building Next Generation Semantic Communication Networks

@article{Chaccour2022LessDM,
  title={Less Data, More Knowledge: Building Next Generation Semantic Communication Networks},
  author={Christina Chaccour and Walid Saad and M{\'e}rouane Debbah and Zhu Han and H. Vincent Poor},
  journal={ArXiv},
  year={2022},
  volume={abs/2211.14343}
}

• Christina Chaccour, W. Saad, +2 authors H. V. Poor
• Published 25 November 2022
• Computer Science
• ArXiv

—Semantic communication is viewed as a revolution- ary paradigm that can potentially transform how we design and operate wireless communication systems. However, despite a recent surge of research activities in this area, remarkably, the research landscape is still limited in at least three ways. First, the very definition of a “semantic communication system” remains ambiguous, and it differs from one work to another. Second, there is a lack of fundamental and scalable frameworks for building next-generation semantic communication networks based on rigorous and well-defined technical foundations. Third, the question of what a “semantic representation” means, and on how this representation can be used to instill meaning, significance, and structure to every information transfer over a wireless network remain unanswered. In this tutorial, we present the first rigorous and holistic vision of an end-to-end semantic communication network that is founded on novel concepts from artificial intelligence (AI), causal reasoning, transfer learning, and minimum description length theory. We first discuss how the design of semantic communication networks requires a move from data- driven and information-driven AI-augmented networks, in which wireless networks remain “tied” to data, towards knowledge- driven and reasoning-driven AI-native networks in which wireless networks are AI-native and can perform versatile logic. We then distinguish the concept of semantic communications from several other approaches that have been conflated with it. For instance, we opine that effectively and efficiently building next-generation semantic communication networks must go beyond: a) creating a new type of encoder and decoder at the transmitter/receiver side, and b) designing a new “AI for wireless” framework in which AI is used to extract some application features or to fine tune a wireless protocol or algorithm. Then, we identify the main tenets that are needed to build an end-to-end semantic communication network. Among those building blocks of a semantic communication networks, we highlight the necessity of creating semantic representations of