Modelling and optimization of multilayer RF passives using coupled neural networks and genetic algorithms

We present the first combined neural network and genetic algorithm based modelling, design and optimization of multilayer passives. The embedded multilayer passives considered are of great interest for RF system-on-package for W-CDMA and C-band applications. An accurate neural network model for multilayer inductors and capacitors is developed using measured results for the frequency range of 1-5 GHz. The neural network model is used to perform sensitivity analysis and derive response surfaces. An innovative technique is then applied in which genetic algorithm based optimization is coupled with neural network modelling of electrical performance. The proposed neuro-genetic algorithm based design promises to minimize the time and cost for multilayer passive design while providing greater accuracy.