A New Hardware-oriented Spiking Neuron Model Based on SET and Its Properties

Abstract

In this paper, a new simple hardware-oriented spiking neuron model is proposed. It is mainly based on one of intrinsic features of single- electron transistor (SET)––Coulomb Oscillation. In fact, Coulomb Oscillation can be considered as current pulses under the gate's voltage, which is similar with the pulses of spiking neurons. Accordingly, the circuit structure of spiking neuron model has been designed. By simulation on the PSPICE, the proposed new model can show different spiking types according to the different input signals and implement information encoding. The model can exhibit the property of synchronization. Unlike traditional neuron model, the proposed new model has the capacity of integrating both spatial and temporal signals. In addition the phase encoding is also involved. So the experiments’ results demonstrate that the model is an implement way of hardware-oriented spiking neuron models or even spiking neural networks (SNN).