ETD EMBARGOED

Machine-Learning-Assisted Synthesis of Active and Passive Matching Circuits and Full-Duplex Antennas for HF/VHF/UHF Applications

Embargoed until 2024-05-25.
Citation

Li, Qianyi. (2022-05). Machine-Learning-Assisted Synthesis of Active and Passive Matching Circuits and Full-Duplex Antennas for HF/VHF/UHF Applications. Theses and Dissertations Collection, University of Idaho Library Digital Collections. https://www.lib.uidaho.edu/digital/etd/items/li_idaho_0089e_12389.html

Title:
Machine-Learning-Assisted Synthesis of Active and Passive Matching Circuits and Full-Duplex Antennas for HF/VHF/UHF Applications
Author:
Li, Qianyi
Date:
2022-05
Embargo Remove Date:
2024-05-25
Keywords:
electrically small antennas in-band full-duplex machine learning non-Foster circuits simultaneous transmit and receive antenna wide bandwidth
Program:
Electrical and Computer Engineering
Subject Category:
Electrical engineering; Electromagnetics
Abstract:

Antennas and corresponding matching circuits for 5G and next-generation communications face many challenges, such as the port isolation between the transmit and receive chains of in-band full-duplex systems, and/or the theoretical bandwidth limitation for electrically small antennas, especially in the HF/VHF bands. Chapter 1 of this thesis discusses these challenges.

This thesis presents a characteristic-mode-based simultaneous transmit and receive (STAR) antenna design for in-band full-duplex (IBFD) systems in Chapter 2. The proposed method utilizes two characteristic modes of a conducting object as transmit and receive modes to achieve high isolation without a complicated self-interference cancellation circuit. The design example in this work is fully-planar, and it has a physical height of 1.6 mm. The measured -10 dB overlapped percentage S11 and S22 bandwidth of this STAR antenna is 2.5% and the measured isolation between the transmit and receive ports is greater than 30 dB over the entire frequency band of operation.

The bandwidth limitation of an electrically-small antenna can be overcome by employing active non-Foster matching circuits. Considering the time of designing stable non-Foster circuits, this dissertation presents a real-time machine-learning-assisted RF circuits synthesis toolbox. In Chapter 3, an application programming interface (API) and a toolbox were developed for automatically generating a massive amount of data samples for machine learning model training. The details of the automatic non-Foster circuits synthesis tool using different machine learning models are introduced in Chapter 4, including the artificial neural network (ANN), k-nearest neighbor (k-NN), and Gaussian process (GP) algorithms. A non-Foster circuit prototype was fabricated and measured to validate the proposed tool. The measured reactance shows a negative reactance-to-frequency slope, which confirms it is a non-Foster circuit. The proposed framework can be expanded to include more design features and other circuit topologies.

Description:
doctoral, Ph.D., Electrical and Computer Engineering -- University of Idaho - College of Graduate Studies, 2022-05
Major Professor:
Shih, Ting-Yen
Committee:
Chakhchoukh, Yacine; Zadehgol, Ata; Ma, Marshall; Law, Joseph
Defense Date:
2022-05
Identifier:
Li_idaho_0089E_12389
Type:
Text
Format Original:
PDF
Format:
record

Contact us about this record

Rights
Rights:
In Copyright - Educational Use Permitted. For more information, please contact University of Idaho Library Special Collections and Archives Department at libspec@uidaho.edu.
Standardized Rights:
http://rightsstatements.org/vocab/InC-EDU/1.0/