Optimization of RF MEMS Metal-Contact Switch using Numerical and Evolutionary Methods

Abstract

This paper presents the design, simulation, and optimization of radio frequency
microelectromechanical system (RF MEMS) metal-contact switch. The novel resistive switch is
simulated and optimized using numerical and evolutionary methods. In this paper,the designed switch
is simulated using genetic algorithm, pattern search, sequential non-linear programming, and quasi
newton method to get the low insertion, high isolation, and better return loss performance. The
designed switchcan be used for reconfigurable antenna circuits up to 10 GHz applications. The switch
offers the best optimization results using a sequential non-linear programming approach both in the
ON and OFF positions, respectively. The pattern search method approximately approaches the
sequential non-linear programming method with less time consumption and the number of iteration.
Further, the artificial neural network is implemented to verify the outcomes of numerical and
evolutionary methods in terms of mean square error and regression. The MEMS switch shows
optimized insertion losses in the range lower than 0.392 dB in ON position and isolation in the range
as high as 106.57 dB in OFF position.