A dynamic examination of the digital circuit implementing the Fitzhugh-Nagumo neuron model with emphasis on low power consumption and high precision.

Neuromorphic computing has got more attention in various tasks during recent years. The main goal of this field is to explore neural functionality in the brain. The studies of spiking neurons and Spiking Neural Networks (SNNs) are vital to understand how brain-inspired neural system work. In this paper, the modified Fitzhugh-Nagumo (FHN) model is proposed based on Coordinate Rotation Digital Computer (CORDIC) algorithm to emulate biological behaviors of the original neuron model. The presented CORDIC method eliminates multipliers by using adder and shifter operations, which provides efficient digital hardware implementation of the FHN model. Error analysis and dynamic assessments confirm that the CORDIC-based FHN model is capable to follow the biological behaviors of the original model with high accuracy. Additionally, to further check the compatibility of the CORDIC-based FHN model with the original model, chaotic behaviors of both models on bifurcation and maximum Lyapunov exponent diagrams are investigated. Considering that the CORDIC-based FHN model has a high compatibility with the original model, its superiority over the original model is the possibility of hardware implementation with low power consumption. To analyze further, two cost functions are defined based on operation frequency, power, and error to confirm the efficiency of the proposed hardware compared to previous studies. As a result of its low power consumption, minimal error rates, and high-frequency capabilities, the proposed hardware demonstrates effectiveness and utility across a range of applications, including the simulation of learning processes in the nervous system that are based on nonlinear and chaotic behaviors.