The simplest and efficient form of Spiking Neuron

Neuromorphic computing is an emerging field that mimics the functionality of human brain to overcome the paradigm of conventional computing. This uses artificial (silicon) neuron as a basic building block in their architectures. There are some neuron models which exactly resembles the biological nature, but complex to design, consumes more hardware to implement, inefficient in computation and some models are very simple and computationally efficient. This paper proposes and discusses the simplest and efficient neuron model for generation of various spiking patterns like tonic spiking, class one excitable and integrator. The proposed model is a reduction model from the Integrate and Fire (IAF) neuron. Izhekivich, IAF and the proposed models are implemented using MATLAB and Simulink. It is observed that the proposed model and IAF model give the same spiking patterns. In addition, the proposed model gives more spike count when compared with the IAF in a given time period. Hence, the proposed one is much faster and simple to implement as it is a first order differential equation with constant slope, and with reduced number of variables. The analog version of the proposed neuron is implemented using GPDK library at 180 nm technology node, for generating tonic spiking, phasic spiking and bursting.