STOCHASTIC RESONANCE AND NOISE-ENHANCED FISHER INFORMATION

Abstract

We discuss the signal estimation that can be finished on a signal buried in generalized Gaussian noise based on the quantized version provided by a summing array of threshold devices. In the estimation, the Fisher information contained in the array output about the input signal is investigated. We show that the Fisher information can be improved as the noise intensity increases in the summing array and that a noise with a thinner tail in its distribution can lead to a better improvement, i.e., the fatter tail in the noise distribution may neutralize the beneficial role of noise. These results prove that the phenomenon of stochastic resonance (SR) or supra-threshold stochastic resonance (SSR) exists based on Fisher information in the summing array of threshold devices for generalized Gaussian noise, and that the noise distribution has an effect on the efficacy of SR or SSR. These above results also extend the applicability of the nonlinear phenomenon of SR or SSR in signal estimation.