GenAI and LLMs for Source and Channel Coding in B5G Networks
Publication Date
10-29-2025
Document Type
Contribution to a Book
Publication Title
Genai and Llms for Beyond 5g Networks
DOI
10.1007/978-3-032-06418-9_9
First Page
197
Last Page
216
Abstract
The evolution of Beyond 5G (B5G) networks demands communication systems that are ultrareliable, low latency, and capable of handling diverse, data-intensive applications. Traditional source and channel coding methods, following Shannon’s separation principle, often fall short in dynamic and heterogeneous wireless environments. This chapter reviews the foundational principles of source and channel coding, highlighting key theoretical insights, and then delves into recent advances in deep learning, Generative AI (GenAI), and Large Language Model (LLM)-based joint source-channel coding techniques. Emphasizing semantic and task-oriented communication, these emerging approaches offer promising solutions for more intelligent, adaptable, and efficient transmission strategies tailored to the new requirements of B5G networks.
Keywords
Additive White Gaussian Noise (AWGN), Beyond 5G (B5G) wireless, Channel Capacity, Entropy, Fading, Generative AI (GenAI), Information Theory, Joint source-channel coding (JSCC), Large Language Model (LLM), Memoryless, Mutual information
Department
Computer Science
Recommended Citation
Jaswanthi Mandalapu, Abhishek Roy, and Navrati Saxena. "GenAI and LLMs for Source and Channel Coding in B5G Networks" Genai and Llms for Beyond 5g Networks (2025): 197-216. https://doi.org/10.1007/978-3-032-06418-9_9
