A bio-inspired single-electron circuit exploiting noises to achieve high temporal fidelity based on the vestibulo-ocular reflex function

摘要

Although deep sub micron microelectronic technologies enable greater degrees of semiconductor device integration, this is coupled with increased degrees of variations in physical sizes of devices (static noises) and increased device sensitivity to internal and external (temporal) noises. Overcoming these two problems would be even more challenging, to circuit designers, as device sizes are shrunk further into the nano scales. On the other hand, electrical circuits that arc designed by mimicking computational structures in living organisms-ncuromorphic circuits ([1], [2]), arc viewed as viable insights on how to handle these noises toward developing efficient processors with such failure-prone devices. In this work, inspired by signal encoding in the vestibule-ocular reflex function {[41), we present a novel architectural approach where deviee fluctuations and temporal noises are actively employed to enhance the performance of neuronal circuits consisting of single-electron devices. We confirmed that employing noises reduces the probability of synchrony among the neurons, leading to well distributed firing events in the network. Consequently, this enhances the fidelity with which neurons can encode high-frequency input signals.