Date of Award

8-2021

Document Type

Dissertation - Campus Access Only

Degree Name

Doctor of Philosophy (PhD)

Degree Discipline

Electrical Engineering

Abstract

The emergence of the fifth-generation wireless communications system (5G) and the exponential increase of the Internet-of-Things (IoT) has led to the increase in demand for scarce spectral resources. The future wireless communication system will be heterogeneous, and radios will co-exist in shared, licensed, and unlicensed bands. Intelligent capabilities are of utmost importance to achieve this level of co-existence, dynamic response to changes in the environment, and efficient resource utilization. This can be achieved through learning by leveraging the availability of historical spectrum data and the capabilities of deep learning. The choice of parameters, algorithm, performance evaluation, and robustness to adversarial attacks is critical to learn appropriately and develop robust models.

In this dissertation, deep learning techniques for learning from radio frequency (RF) data to achieve various wireless communication domain objectives are proposed. In addition, a testbed for data collection using LabVIEW software and universal software radio peripheral (USRP) devices has been built. A novel approach to data selection for training and evaluation of deep learning models is proposed considering essential scenarios in wireless communication systems such as spectrum monitoring, device identication, and automatic modulation classication. The effect of signal to-noise ratio (SNR) in model performance is evaluated. Furthermore, a study of adversarial attacks to establish their impact on wireless communication systems and methods to prevent it is presented. Experimental results demonstrate the effectiveness of the proposed deep learning approach and show the effect of adversarial machine learning in wireless communications.

Committee Chair/Advisor

Lijun Qian

Committee Member

John Fuller

Committee Member

Xiangfang Li

Committee Member

Pamela Obiomon

Committee Member

Huang Lei

Publisher

Prairie View A&M University

Rights

© 2021 Prairie View A & M University

Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

Date of Digitization

11/07/2024

Contributing Institution

John B Coleman Library

City of Publication

Prairie View

MIME Type

Application/PDF

Download

Share

COinS