Optimal weighted suprathreshold stochastic resonance with multigroup saturating sensors

Abstract

Suprathreshold stochastic resonance (SSR) describes a noise-enhanced effect that occurs, not in a single element, but rather in an array of nonlinear elements when the signal is no longer subthreshold. Within the context of SSR, we investigate the optimization problem of signal recovery through an array of saturating sensors where the response of each element can be optimally weighted prior to summation, with a performance measure of mean square error (MSE). We consider groups of sensors. Individual sensors within each group have identical parameters, but each group has distinct parameters. We find that optimally weighting the sensor responses provides a lower MSE in comparison with the unweighted case for weak and moderate noise intensities. Moreover, as the slope parameter of the nonlinear sensors increases, the MSE superiority of the optimally weighted array shows a peak, and then tends to a fixed value. These results indicate that SSR with optimal weights, as a general mechanism of enhancement by noise, is of potential interest to signal recovery.