Adaptive Deep Learning in Physical Layer Applications

Author

Ali OwfiFollow

Date of Award

12-2025

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Electrical and Computer Engineering

Committee Chair/Advisor

Dr. Fatemeh Afghah

Committee Member

Dr. Linke Guo

Committee Member

Dr. Xiaoyong Yuan

Committee Member

Dr. Vuk Marojevic

Abstract

Traditionally, signal processing models in communication systems have been designed based on solid foundations in statistics and information theory, often assuming linearity and optimizing for simplified models. However, real-world communication systems exhibit numerous imperfections and non-linearities that traditional linear models struggle to capture accurately. Deep Learning (DL)-based approaches, unconstrained by rigid mathematical models, have shown promise in optimizing system performance by accommodating specific hardware configurations and dynamic channel conditions as an alternative to the traditional methods. Despite all the recent research efforts on DL-based methods for physical layer applications, DL models have still not been widely applied to physical layer applications, and yet remain to be commercialized. This gap between research and deployment can be attributed to various challenges, including computational complexity, lack of standardization, integration hurdles, reliability concerns, regulatory issues, and cost considerations. But more importantly, there are still issues with current DL-based physical layer designs that prevent their widespread practical use. A significant limitation of many proposed DL-based physical layer designs is their lack of adaptability in dynamic situations. Current designs often rely on stationary assumptions and are trained and tested using datasets collected for specific channel conditions. However, real-world communication systems are rarely stationary, and models optimized on static datasets may not generalize well to dynamic communication scenarios. This dissertation aims to address this challenge by proposing and evaluating advanced DL methodologies, such as meta-learning, incremental learning, and domain adaptation, that enhance adaptability of DL-based physical layer models across dynamic scenarios.

Recommended Citation

Owfi, Ali, "Adaptive Deep Learning in Physical Layer Applications" (2025). All Dissertations. 4139.
https://open.clemson.edu/all_dissertations/4139

Author ORCID Identifier

0009-0009-7237-9594