Design of radiofrequency wireless sensing devices based on resonance techniques

Microwave sensing based on resonance techniques has become a promising solution to address the growing demand for portable, low-cost, and stand-alone diagnostic systems. By exploiting the interaction between electromagnetic (EM) fields and the material under test (MUT), these sensors provide accurate and reliable measurements across diverse application domains. In this framework, the integration of resonant structures with machine learning (ML) algorithms enhances sensing performance by enabling adaptive, real-time data processing and improved handling of complex and variable conditions. This Thesis presents the design and experimental validation of innovative microwave sensors developed for precision agriculture and biomedical applications. The main objective is to ensure high sensing accuracy while achieving full portability, thereby eliminating the need for bulky and expensive laboratory radiofrequency (RF) instrumentation typically required in conventional microwave measurement setups. Two distinct sensing approaches are investigated. The first focuses on in-situ monitoring of tree trunk moisture content. A compact, low-profile self-oscillating antenna-based sensor is developed, introducing a novel strategy based on analyzing the steady-state operating regimes of the oscillator. These regimes vary according to changes in antenna loading caused by moisture variations. By monitoring only the DC power consumption associated with each regime, the system provides a robust and simplified alternative to traditional RF measurement techniques. The second approach addresses biomedical sensing, specifically non-invasive skin hydration assessment. Electrode-less resonators operating at different frequencies are designed to achieve controlled penetration depths. In particular, a high-resolution resonator is optimized to selectively characterize the most superficial, non-vascularized skin layer, enabling the detection of microscopic fissures associated with atopic dermatitis. Overall, the proposed systems demonstrate that resonance-based microwave sensing, combined with ML techniques, can enable compact, autonomous, and adaptable diagnostic platforms suitable for real-world agricultural and biomedical applications.