Efficient unsupervised synthesis and prototype-augmented modeling of radio-frequency structures - Project - Bridge of Knowledge

Your account

login

Search

Efficient unsupervised synthesis and prototype-augmented modeling of radio-frequency structures

Topology synthesis is a significant challenge related to design of modern radio-frequency components. Conventional design scheme involves pre-determination of the structure topology (based on empirical approaches, or available theory) followed by its manual adjustments oriented towards achieving the desired functionality. The role of numerical optimization in the outlined workflow (if any) is limited to parametric tuning of the synthesized, and hence topologically restricted, structure. The heuristic approach to the problem seems suitable having in mind that component synthesis is an NP complex problem. On the other hand, experience-driven methods are often inefficient when complex requirements are to be fulfilled. Alternative techniques, oriented towards unsupervised synthesis of structures using numerical methods, are prohibitively expensive due to tremendous number of EM simulations required for generation of the circuit. Consequently, their practical applications are limited to rather simple components with low number of parameters. The goal of the project is development of reliable tools for automated generation, optimization, yield-oriented tuning, and prototype-enhanced design of radio-frequency components. The key technologies that will be used include machine learning, efficient optimization tools, surrogate-assisted statistical analysis, and additive manufacturing techniques. The latter are essential for cheap, rapid fabrication of components and their measurements. The obtained experimental data will be used as a feedback for enhancing the data-driven analysis, and modeling of components. This “prototype-rich” framework will be essential for shortening components time-to-market cycles and increasing their immunity to manufacturing tolerances. The proposed project will leverage the potential of advanced modeling methods and cheap additive manufacturing in order to reduce engineering insight in the development of contemporary radio-frequency components. It is expected that successful realization of the project will not only contribute to development of the microwave/antenna engineering, but also other areas where bias-free synthesis of unconventional components is of importance. From this perspective, the project is important for the Gdansk University of Technology.

Details

Financial Program Name:
OPUS
Organization:
Narodowe Centrum Nauki (NCN) (National Science Centre)
Agreement:
UMO-2021/43/B/ST7/01856 z dnia 2022-07-04
Realisation period:
unknown - unknown
Project manager:
dr hab. inż. Adrian Bekasiewicz
Realised in:
Department of Microelectronic Systems
Request type:
National Research Programmes
Domestic:
Domestic project
Verified by:
Gdańsk University of Technology

Papers associated with that project

Filters

total: 7

  • Category

  • Year

  • Options

clear Chosen catalog filters disabled

Catalog Projects

Year 2024

seen 340 times