CNNS for noise generation in dithered transducers

A real time application of noise-added techniques requires the use of a suitable noise generator. Several solutions have been proposed but very often they were too complex and free of applicability. In this paper the possibility of using a Cellular Neural Network as a noise generator is investigated. Indeed, it is well known that CNNs can have very complex dynamics and are analog devices that are capable of working on line as signal generators. In particular a CNN implementing the Chua system generating both Gaussian and uniform white noise is discussed, and suitable techniques for CNN parameter estimation are presented. An analog implementation of the investigated system is proposed.