A Nanoscale Bistable Resistor for an Oscillatory Neural Network

Area of Science:

• Electronics
• Neuroscience
• Computer Engineering

Background:

• Oscillatory neural networks (ONNs) mimic brain neuron interactions.
• Traditional ONNs often lack CMOS compatibility and power efficiency.

Purpose of the Study:

• To demonstrate a fully CMOS-based oscillatory neural network using novel biristor oscillators.
• To investigate the coupled oscillation characteristics and potential applications of these birillators.

Main Methods:

• Fabrication of a CMOS-based bistable resistor (biristor) oscillator.
• Experimental investigation of two coupled birillators, observing 180° phase synchronization.
• Implementation of edge detection and vertex coloring using phase differences in the ONN.

Main Results:

• The biristor oscillator generates spike-like oscillating voltage signals via a single transistor latch.
• Synchronized coupled birillators exhibit a 180° phase difference, suitable for ONN applications.
• Successful demonstration of edge detection and vertex coloring using the proposed ONN.

Conclusions:

• Fully CMOS-based biristor oscillators offer a low-power, highly compatible, and compact solution for ONNs.
• The synchronized coupled birillator system shows promise for neuromorphic computing and signal processing.
• This work advances the development of efficient hardware for artificial neural networks.