Title:
Machine Learning for Automating Millimeter-Wave Electromagnetic Designs
Machine Learning for Automating Millimeter-Wave Electromagnetic Designs
| dc.contributor.advisor | Peterson, Andrew F. | |
| dc.contributor.author | Nguyen, Huy Thong | |
| dc.contributor.committeeMember | Cressler, John | |
| dc.contributor.committeeMember | Klein, Benjamin | |
| dc.contributor.committeeMember | Durgin, Gregory David | |
| dc.contributor.committeeMember | Nguyen, Thanh | |
| dc.contributor.department | Electrical and Computer Engineering | |
| dc.date.accessioned | 2021-01-11T17:12:16Z | |
| dc.date.available | 2021-01-11T17:12:16Z | |
| dc.date.created | 2020-12 | |
| dc.date.issued | 2020-12-01 | |
| dc.date.submitted | December 2020 | |
| dc.date.updated | 2021-01-11T17:12:16Z | |
| dc.description.abstract | This dissertation proposes a general class of solution and Machine Learning (ML) techniques to support the designs of several critical Electromagnetic (EM) structures at mm-wave frequency ranges, for which the main applications are to directly address the emerging power and efficiency challenges of the next generation of wireless communication. Starting from the coupled line theory, we theoretically propose a common solution for Impedance Transforming Baluns, Power Combiners, Out-Phasing circuits, and Doherty networks, which we refer to as the BCOD structure. The main contribution of this dissertation is to develop Machine Learning (ML) techniques that, within the computational time of seconds, can fully automate EM designs of the BCOD network that achieves the lowest metal loss, in which we demonstrated the proposed ML approaches on various on-chip metal stacks for a wide range of electrical specifications. Notably, optimizing for the lowest metal loss is a challenging problem, and to the best of our knowledge, we are not aware of any prior techniques that can systematically do so. Importantly, the application of our proposed ML approaches can go beyond the task of automating EM structures. Serving as a new tool for bigger optimization loops, the ML techniques can theoretically answer several challenging, abstract, and high-level questions for mm-wave designs, such as the calculation of the optimum transistor sizes, or the derivation of the rule of thumb between device sizes and mm-wave frequencies. | |
| dc.description.degree | Ph.D. | |
| dc.format.mimetype | application/pdf | |
| dc.identifier.uri | http://hdl.handle.net/1853/64157 | |
| dc.language.iso | en_US | |
| dc.publisher | Georgia Institute of Technology | |
| dc.subject | Automation, Electromagnetics, Machine Learning, Millimeter-Wave | |
| dc.title | Machine Learning for Automating Millimeter-Wave Electromagnetic Designs | |
| dc.type | Text | |
| dc.type.genre | Dissertation | |
| dspace.entity.type | Publication | |
| local.contributor.advisor | Peterson, Andrew F. | |
| local.contributor.corporatename | School of Electrical and Computer Engineering | |
| local.contributor.corporatename | College of Engineering | |
| relation.isAdvisorOfPublication | 080ab541-0779-41ab-ae29-90761a9f6649 | |
| relation.isOrgUnitOfPublication | 5b7adef2-447c-4270-b9fc-846bd76f80f2 | |
| relation.isOrgUnitOfPublication | 7c022d60-21d5-497c-b552-95e489a06569 | |
| thesis.degree.level | Doctoral |