DESIGN AND DEVELOPMENT OF A NOVEL MICROWAVE SENSOR FOR BIOCHEMICAL DETECTION

A non-invasive microwave sensor with a Square Split Ring Resonator (S-SRR) for sensing applications is proposed in this paper.With the rapid advancement of sensing systems and emerging technologies, there is a high demand for efficient and compact microwave sensor designs.This study presents an innovative approach to enhancing the performance of microwave sensors through the integration of an S-SRR structure, achieving an operating frequency of 2.675 GHz with an S-parameter magnitude of -25.43 dB. The proposed sensor demonstrates high sensitivity in detecting the dielectric constants of different analytes, making it suitable for various material characterization applications. To further enhance the accuracy of dielectric constant estimation, machine learning-based regression algorithms have been employed to establish a model describing the correlation between the dielectric constant and the resonant frequency. Multiple regression models, such asRandom Forest regression, Polynomial and Linear regressions, have been systematically analyzed to determine the most accurate predictive model.Among these, Polynomial Regression (Degree 2) emerged as the top-performing model, achieving the lowest Root Mean Squared Error (RMSE) and the highest R score, indicating superior accuracy in dielectric constant estimation.The derived empirical equations from the regression models enable precise forecasting of unknown analyte's dielectric constants, thereby facilitating rapid, non-destructive, and highly precise material characterization.This research contributes to the advancement of microwave-based sensing by integrating machine learning techniques for improved analytical performance, providing a scalable and efficient solution for real-world sensing applications.