Modeling techniques are crucial in designing microwave devices and help to gain a comprehensive understanding of their operation. In recent years, artificial neural networks (ANNs) have proven successful in modeling the frequency-dependent behavior of various microwave devices such as antennas, sensors, and transistors. ANNs are capable of learning from device behavior through a defined training process and can be used as a fast and accurate model for the description of the device under test. In addition to ANN modeling, the Lorentzian fitting procedure is often used to accurately determine the resonant parameters of microwave resonators. The accurate determination of the resonant frequency and quality factor is desirable regardless of the type of resonator or its application (e.g., filter, sensor). Literature reports suggest that the most accurate estimation of these resonant parameters is obtained by measuring the scattering parameters of the device under test using a vector network analyzer (VNA), followed by fitting a Lorentzian function. This approach compensates for resonator non-idealities, which may affect the accurate determination of both resonant frequency and Q-factor.

In a recent study, the University of Messina (Italy), in collaboration with the University of Niš (Serbia), has developed a new and more reliable modeling tool for studying microwave resonators by combining the Lorentzian fitting method and the ANN-based modeling approach. This approach combines the benefits of both methods and has been successfully applied to model a microwave sensor for relative humidity measurements.

The research, which was co-authored by Prof. Zlatica Marinković from the University of Niš, Dr. Giovanni Gugliandolo, Prof. Giuseppe Campobello, Prof. Giovanni Crupi, and Prof. Nicola Donato (Res4Net member) from the University of Messina, was presented at the 2022 IEEE International Conference on Metrology for eXended Reality, Artificial Intelligence, and Neural Engineering (IEEE MetroXRAINE 2022) and was recognized with the “Best Paper Presented by a Young Researcher” award.

For a more detailed description of the research, please refer to the conference proceeding webpage.