Complex permittivity characterization of textile materials by means of surrogate modelling

In this paper, we perform electromagnetic characterization of textile materials by means of fitting simulated and measured textile antenna performances. The process is automated by using an optimization method based on kriging surrogate models. More in particular, we apply expected improvement (EI) as implemented in a flexible research platform called the Surrogate Modelling (SUMO) Toolbox. Previously, kriging surrogate models were used for EM device optimization, in which the surrogate model is then optimized instead of an expensive simulation. Here, we convert the inverse problem to a forward optimization problem by minimizing an error function between the simulated and measured data. Results of two different textile materials serving as antenna substrate are presented and compared with the results obtained by a direct measurement method in which the electromagnetic properties are determined based on microstrip line measurements.