Applying stochastic signal quantization theory to the robust digitization of noisy analog signals

Abstract

Suprathreshold stochastic resonance is a variant of stochastic resonance that has been shown to occur in parallel arrays of independently noisy, but otherwise identical, binary threshold devices. It can be described as a form of stochastic signal quantization that utilizes independently random noise sources to digitize an analog signal. This paper outlines a generalization of this effect, and discusses several sensor applications in which it can occur, such as analog-to-digital converter circuits, distributed sensor networks and the reduction of in-band noise via coherent integration in radar, sonar and sodar systems. All of these scenarios can be modeled using a framework called a 'stochastic pooling network'. © 2009 Springer-Verlag Berlin Heidelberg.