Modeling and Optimization of Ohmic Series RF MEMS Switches by using Neural Networks
Conference: GeMiC 2014 - German Microwave Conference
03/10/2014 - 03/12/2014 at Aachen, Germany
Proceedings: ITG-Fb. 246: GeMiC 2014
Pages: 4Language: englishTyp: PDFPersonal VDE Members are entitled to a 10% discount on this title
Milijic, Marija; Marinkovic, Zlatica; Pronic-Rancic, Olivera; Markovic, Vera (University of Niš, Faculty of Electronic Engineering, Aleksandra Medvedeva 14, 18000 Niš, Serbia)
Kim, Teayoung; Vietzorreck, Larissa (TU Muenchen, Lehrstuhl für Hochfrequenztechnik, Arcisstr. 21, 80333 Munich, Germany)
In this paper, artificial neural networks (ANNs) are proposed to be used as a very efficient simulation and optimization tool to replace time consuming full-wave simulations of electrical characteristics of an ohmic series RF MEMS switches in coplanar technology. ANNs are trained to determine the scattering parameters of a switch for given values of lateral dimensions of the switch and the frequency. Not only the bridge lateral dimensions but also lateral dimensions of the gap between bridge and ground are considered. By using the proposed ANN based model, the time needed for device design is significantly reduced, especially when optimizations of the dimensions are required in order to meet the specified criteria.