An Analog Implementation of FitzHugh-Nagumo Neuron Model for Spiking Neural Networks

Abstract

A low power analog implementation of FitzHugh-Nagumo (FHN) neuron model is presented in this paper for large scale spiking neural network and neuromorphic algorithm realization. The FHN neuron model is designed using $\log $-domain low pass filters and translinear multipliers to emulate voltage-like variable with cubic non-linearity and a recovery variable. Various spiking behaviors observed in biological neurons are demonstrated in simulation results. The neuron model was designed in 45 nm CMOS process which has 1.6 nW and 40 nW power consumption at rest and for a single spiking event respectively.