A Low-Energy Refractory Spike Frequency Adaptive LIF Neuron Model
Arnab Deb, David Selvakumar · 2024
This paper presents a Leaky Integrate-and-Fire (LIF) neuron model, featuring a refractory period control unit and spike frequency adaptation, developed using 180nm CMOS technology. The goal is to realize the intricate characteristics of biological neurons closely. The LIF neuron circuit accommodates inputs from both excitatory and inhibitory synapses, responding accordingly. To ensure its reliability, simulation tests are conducted, including recording its response to a ramp voltage-modulating the spike frequency adaptation. Tests also examined the responses to input pulse waveforms with and without a refractory unit, revealing a refractory period of 1.5µs. Circuit-level simulations showed an energy efficiency of 110fJ/spike with a static power dissipation of 150nW. Additionally, a neuron network with two similar LIF neurons and an ideal neuron interconnect in a unidirectional setup is formed, testing them with input waveforms to observe spike generation and propagation with a refractory unit.