Digital multiplierless realization of two coupled biological Morris-Lecar neuron model

Abstract

Modeling and implementation of biological neural networks are significant objectives of the neuromorphic research field. In this field, neuronal synchronization plays a significant role in the processing of biological information. This paper presents a set of piecewise linear (MLPWL1) and multiplierless piecewise linear (MLPWL2) neuron models, which mimic behaviors of different types of neurons, similar to the biological behavior of conductance-based neurons. Both simulations and a low-cost digital implementation are carried out to compare the proposed models to a single ML neuron and two coupled ML neurons, demonstrating the required range of dynamics with a more efficient implementation. Hardware implementations on a field-programmable gate array (FPGA) show that the modified models mimic the biological behavior of different types of neurons with higher performance and significantly lower implementation costs compared to the previous realizations of the ML model. The mean normalized root mean square errors (NRMSEs) of the MLPWL1 and MLPWL2 models are 3.70% and 4.89%, respectively, as compared to the original ML model.