2016 Volume 7 Issue 2 Pages 164-175
Stochastic resonance (SR) is a phenomenon in which dynamic noise is effectively used to drive a system with subthreshold input signals. In the classical model of SR induced in a network, each constituting unit must be delivered noise from independent external sources. Recently, a new model of SR has been proposed, where internal noise is exploited as the solution to avoid the burden of generating independent external noise for each unit. In this study, we employ a network of FitzHugh-Nagumo neurons as a candidate of the new model of SR system using internal noise. The network is formed as a circular system where all connections between neurons strictly consist of self-connections or connections propagating into a unique direction. Hence, each neuron receives stimuli from its four predecessors within the circular arrangement, from its own output, as well as from a unique input node. An input signal of an amplitude that is smaller than the threshold of individual neurons is provided to the network. Owing to a process of SR that occurs within the network and that is sustained by internally generated noise, the subthreshold signal is detected and amplified and delivered to the network output node. The frequency characteristics of the network in terms of its operational bandwidth is established.
