Date of Award

12-1-2022

Language

English

Document Type

Master's Thesis

Degree Name

Master of Science (MS)

College/School/Department

Department of Electrical and Computer Engineering

Content Description

1 online resource (xi, 74 pages) : illustrations (some color)

Dissertation/Thesis Chair

Dola Saha

Committee Members

Aveek Dutta, Hany Elgala

Keywords

Deep learning for Wireless Communication, Interpretable AI for Wireless, Terahertz band Communication, Wireless Communication, Orthogonal frequency division multiplexing, Deep learning (Machine learning), Wireless communication systems

Subject Categories

Computer Engineering | Electrical and Electronics

Abstract

Wireless receiver design for OFDM systems is well investigated with classical signal processing tools, which lack the capacity to extract intrinsic channel effects in received signal and lead to high decoding error in receiver. Current deep learning techniques have shown improvement in such cases. But these models are mostly being developed as black box without any anchor to the theory of wireless signal propagation, which leads to surface level information gain and lacks generalizability. We propose deep learning models where the hyperparameters and learning objectives are derived from domain knowledge of wireless signal propagation. These models not only increase the quality of channel estimation and equalization due to their capability to learn precise nonlinear functions, but take care of the primary drawbacks of popular deep learning models such as impractical data need and frequent retraining as well. On the other hand, for emerging spectra such as Terahertz band, that are in investigative stages currently, can benefit from such models as well. Since, the models are developed exploiting the fundamental features of wireless signal propagation, they help explore the channel and signal features of Terahertz band to develop an efficient and practical channel estimation technique and subsequently an end to end physical layer receiver design.

Download

Share

COinS