Bio-inspired single-electron circuit architectures exploiting thermal noises and device fluctuations to enhance signal transmission fidelity

This paper discusses the implications of noises in a pulse-density modulation single-electron circuit based on vestibulo-ocular reflex model. The proposed circuit consists of an ensemble of single-electron integrate-and-fire neurons that encode the input voltage into pulses whose temporal density is proportional to the amplitude of the input. We confirmed that static noises (heterogeneity in circuit parameters) and dynamic noises (random firing) introduced into the network indeed played an important role in improving the fidelity with which the neurons could encode signals with input frequencies higher than the intrinsic response frequencies of single neurons or a network of neurons without noises. Through Monte-Carlo based computer simulations, we demonstrated that noises could enhance the fidelity with which the network could correctly encode signals with high input frequencies: a noisy network could operate over a wider input range than a single neuron or a network of homogeneous neurons.