An Energy-Efficient Ge-Based Leaky Integrate and Fire Neuron: Proposal and Analysis

Abstract

In this paper, we propose an energy-efficient Ge-based Leaky Integrate and Fire (LIF) neuron and analyze it using a well calibrated 2D simulation model. The proposed neuron can directly receive the incoming voltage spikes and avoid the energy dissipation in generating a summed potential. The incoming voltage spikes lead to accumulation of holes in the channel, leading to lowering of the potential barrier and an increase in current. A firing and subsequent reset circuitry are triggered when the current reaches a predefined threshold. The smaller bandgap with dominant direct tunneling of Ge allows the device to operate at a lower voltage level. The energy consumption per spike in the proposed implementation is 0.07fJ, which is lower than LIF neuron implementations (experimental or simulated) reported in the literature. Power consumed by the reset circuitry can also be reduced due to a lower drain voltage required in the proposed device.