Characterization of Conductor-Backed Dielectric Materials With Genetic Algorithms and Free Space Methods

This work demonstrates the use of genetic algorithms (GA) to perform electromagnetic characterization of dielectric materials. The characterization process entails that the GA generates candidate solutions of the permittivity; these solutions are then used to compute the Fresnel reflection coefficient. The permittivity value which produces the reflection coefficient that best matches the measured reflection coefficient is considered the best solution. Experimental results on a Plexiglas sample are presented. Accompanying reliability analysis of the cost function also shows the method is accurate and stable.