Key Study Cases of Generative AI Applications to Communications Systems

In this chapter, applications of generative AI (GenAI) in communication systems are studied. First, we review the main use cases of key generative models in communication networks. Then, we focus on denoising diffusion models, and with in-depth technical and practical details, we show how diffusion models can help improve the performance of communication systems. Specifically, we show that the generative prior, together with the denoising process of diffusion models helps improve the receiving performance under extreme channel conditions. Intuitively speaking, the idea behind diffusion models is to decompose data generation process over “denoising” steps, and gradually generate samples out of noise. While this fundamentally differs from conventional generative models, the mechanism highlights a strong relation to the overall goal of a communication system, that is, to denoise and reconstruct information signals out of noisy ones. This has motivated us to mostly focus on diffusion generative models in this chapter. The chapter offers a comprehensive overview of the theoretical foundations, to better understand the underlying concepts of diffusion models. Numerical results and comparisons with respect to generative versus discriminative approach are also discussed. We also provide experimental methodologies and practical implications to guide readers to do further experiments in the realm of GenAI. © 2026 by The Institute of Electrical and Electronics Engineers, Inc. All rights reserved.