Stochastic resonance in an array of integrate-and-fire neurons with threshold

摘要

We investigate the phenomenon of stochastic resonance (SR) in parallel integrate-and fire neuronal arrays with threshold driven by additive noise or signal-dependent noise (SDN) and a noisy input signal. SR occurs in this system. Whether the system is subject to the additive noise or SDN, the input noise eta(t) weakens the performance of SR but the array size N and signal parameter I-1 promote the performance of SR. Signal parameter I-0 promotes the performance of SR for the additive noise, but the peak values of the output signal-to-noise ratio (SNRout) first decrease, then increase as I-0 increases for the SDN. Moreover, when N tends to infinity, for the SDN, the curve of SNRout first increases and then decreases, however, for the additive noise, the curve of SNRout increases to reach a plain. By comparing system performance with the additive noise to one with SDN, we also find that the information transmission of a periodic signal with SDN is significantly better than one with the additive noise in limited array size N.