Computational Extraction of Dielectric Properties from Transmission and/or Reflection Coefficients: A Survey

Abstract

This survey examines the computational methodologies for extracting dielectric properties from transmission and/or reflection coefficients. The overview of conventional measuring techniques – free space measurement, transmission line, resonant methods – and the conversion of the scattering parameters method to dielectric properties such as analytical, numerical analysis, and machine learning techniques, are being explored briefly. Each method has advantages and disadvantages, such as the practicality of the sample size, high-frequency applications, and measurement conditions. The Nicholson-Ross-Weir, National Institute of Standards and Technology, and non-iterative methods provide a straightforward computational extraction technique of dielectric properties. Electromagnetic field analysis and root-finding algorithms improve computational accuracy and stability. Large datasets with varying degrees of complexity can be handled by artificial neural networks, deep neural networks, and neuro-fuzzy networks from machine learning models. This survey provides a computational framework through these various approaches while offering insights into their practicality and effectiveness in characterizing material properties.