Generative Artificial Intelligence Models for Emerging Communication Systems: Fundamentals and Challenges

Zhiyuan Peng, Yuchen Liu, Gaolei Li, Zhaohui Yang, Mingzhe Chen, Dongkuan Xu, Xingqin Lin

Abstract: The evolution of generative artificial intelligence (GenAI) marks a pivotal shift in the potential reshaping of technological landscapes. Wireless networks, bolstered by the advent of advanced intelligent technologies, present a promising domain for leveraging GenAI, which could revolutionize the current networking design and communication paradigms. Extensive research has reviewed large language models, a significant yet specialized facet of GenAI, and explored their integration into networks. This article offers a broad introduction to GenAI and delves into its applications within various emerging communication technologies. We start by providing an overview of representative GenAI models, including variational autoencoder, Transformer, and diffusion models, elucidating their foundational principles. Subsequently, we spotlight their emerging applications in advanced communication systems such as digital twins, integrated sensing and communication, and semantic communication. We also underscore crucial challenges and practical considerations, encompassing aspects like data quality, real-time processing capabilities, privacy risks, and security implications. Furthermore, a prospective outlook on research opportunities is taken to surmount these challenges, thereby unlocking the full potential of GenAI in future communication systems.

External IDs:dblp:journals/cm/PengLLYCXL25