Microwave Resonator for Humidity Detection Applications: A Comparative Analysis between ANNs and Lorentzian Fitting Method

This work focuses on a comparative study between the artificial neural networks (ANNs) and the Lorentzian fitting method, which are applied for modeling of the resonant characteristics of a humidity sensor. The device under test (DUT) is a microwave microstrip resonator, developed on FR4 substrate, with a Ag@a-Fe2O3 nanostructured layer deposited on the gap, aimed at humidity sensing applications. To determine the frequency-dependent behavior of the tested device, the reflection coefficient (Γ) is measured from 3.4 GHz up to 5.6 GHz. It is achieved that each analyzed approach has pros and cons, depending on the given application requirements.