RF-LNA circuit synthesis by genetic algorithm-specified artificial neural network

Author

Dumesnil, Etienne ; Nabki, Frederic ; Boukadoum, Mounir

Author_Institution

Dept. of Comput. Sci., Univ. of Quebec at Montreal, Montreal, QC, Canada

fYear

2014

Firstpage

758

Lastpage

761

Abstract

A genetic algorithm (GA) was used to determine the optimal architecture and input parameters of a feed-forward artificial neural network (ANN), the purpose of which was to synthesize a radio-frequency, low noise amplifier (RF-LNA) circuit. The parameters (chromosomes) processed by the GA included: i) the LNA performance specifications and design constraints; ii) the type of ANN to use multi-layer perceptron (MLP) or radial-basis function (RBF) network; iii) the ANN parameters to set. For two different sets of design parameters, the input/output matching network components and transistor geometries, the GA found ANN solutions capable of predicting their values with success rates above 99 %.

Keywords

genetic algorithms; low noise amplifiers; multilayer perceptrons; neural nets; radial basis function networks; radiofrequency amplifiers; ANN; GA; MLP; RBF network; RF-LNA circuit synthesis; feed-forward artificial neural network; genetic algorithm; matching network components; multilayer perceptron; radial-basis function; radiofrequency low noise amplifier; transistor geometry; Artificial neural networks; Biological cells; Genetic algorithms; Impedance matching; Neurons; Training; Transistors; Synthesis; artificial neural network; genetic algorithm; multilayer perceptron; radial basis function; radiofrequency low Noise Amplifier;

fLanguage

English

Publisher

ieee

Conference_Titel

Electronics, Circuits and Systems (ICECS), 2014 21st IEEE International Conference on

Type

conf

DOI

10.1109/ICECS.2014.7050096

Filename

7050096