A CMOS implementation of FitzHugh-Nagumo neuron model

Abstract

A CMOS circuit is proposed that emulates FitzHugh-Nagumo's differential equations using OTAs, diode connected MOSFETs and capacitors. These equations model the fundamental behavior of biological neuron cells. Fitz-Hugh-Nagumo's model is characterized by two threshold values. If the input to the neuron is between the two thresholds the output yields a sequence of firing pulses, if the input is outside this range, no output is observed. The resulting circuit due to the (voltage) programmability of the OTA allows one to easily vary parameters. Thus a large family of solutions can be obtained including the Van der Pol's equation. Experimental results from a CMOS prototype are given that show the suitability of the technique used, and their potential for biological CMOS system emulation.