HEMT and HBT small-signal model optimization using a genetic algorithm

This paper describes the application of a genetic algorithm to the wideband optimization of the small-signal models of HEMTs and HBTs. The use of a genetic algorithm allows us to overcome some of the difficulties that traditional, gradient-based optimizers encounter in the optimization of complex models such as those used for microwave and millimeter-wave devices. In particular, the need for carefully choosing an initial solution as a starting point for the gradient search disappears altogether. The algorithm was applied to AlGaAs/InGaAs/GaAs PHEMTs, InAlAs/InGaAs/InP HEMTs, GaInP/GaAs and AlGaAs/GaAs HBTs. The paper also deals with the effect of a number of parameters on the computation time and model accuracy, and gives some general guidelines as to how to tailor an efficient optimization algorithm.