Signal reconstruction via noise through a system of parallel threshold nonlinearities, in

We present an analysis of the exploitation of noise for sig-nal reconstruction by an array of nonlinear threshold-based devices. This phenomenon has been described as a form of stochastic resonance known as suprathreshold stochastic resonance. It occurs when all devices in an array of size N have identical thresholds and are subject to independent additive noise. The original work showed that the mutual information between the input and output of the array has a maximum for a nonzero value of noise intensity, for a ran-dom input signal. In this paper we extend the results on this phenomenon to the case of Laplacian signal and noise prob-ability densities, and show conditions exist under which it is optimal. 1