Artificial neural network in modelling of voltage controlled oscillators with jitter

Abstract

The use of artificial neural networks in modeling of oscillators with jitter is proposed in this paper. A new method of simulating oscillators with jitter is developed in which an artificial neural network is used to control the signal generator (i.e. oscillator with adjustable parameters: frequency, amplitude, and jitter). A simple and effective way to generate oscillator output with jitter is proposed. This method ensures a simple, robust, fast and accurate modelling of the oscillators with jitter. The method also yields reasonable results for situations not specified in the learning data set because of the generalization property of artificial neural networks and it is suitable for modelling based on measurements. The models of the GaAs voltage controlled differential ring oscillators for circuit simulation software are developed using the proposed method.