Determination of complex permittivity with neural networks and FDTD modeling



A simple novel cavity-independent method of determination of dielectric properties of arbitrarily shaped materials is presented. Complex permittivity is reconstructed using a neural networking procedure matching the measured and FDTD-modeled frequency characteristics of the reflection coefficient. High accuracy and practical suitability are demonstrated through numerical testing and determination of dielectric properties of fresh and saline water at 915 MHz. © 2004 Wiley Periodicals, Inc. Microwave Opt Technol Lett 40: 183–188, 2004; Published online in Wiley InterScience ( DOI 10.1002/mop.11323