Large-signal physical modeling of active microwave devices using an adaptive real-coded genetic algorithm

A novel large-signal physical modeling approach for active microwave devices using an adaptive genetic algorithm (GA) is presented. The proposed technique solves Poisson´s equation in a two-dimensional quasi-static full hydrodynamic large-signal physical model using a real-coded GA. The proposed technique is validated by simulating a submicrometer MESFET. The physical insight of the problem and being able to develop an accurate mathematical model were the keys to conceiving the new algorithm. The proposed algorithm is compared to different algorithms showing excellent accuracy as well as a reasonable speed of convergence. In addition, the effect of various GA parameters on the algorithm behavior is studied. This is a first step toward applying genetic algorithms to Maxwell´s equations in conjunction with the full hydrodynamic model for global modeling of active microwave devices with the aim of developing a robust and fast simulation approach.