An Adiabatic Regenerative Capacitive Artificial Neuron

Abstract

In recent years, RRAM technology has been actively developed as a means of reducing power dissipation and area in a host of circuits, most notably artificial neuron synapses. However, further reduction in energy consumption may be possible by transitioning to capacitive synapses and combining them with adiabatic technique. In this work, we present and analyse the function and power dissipation of an artificial neuron with capacitive synapses where the synaptic tree is fed by a regenerative clock. Whilst the weights are fixed in this case, developments into memcapacitor technology offer the promise of tuneability in the future. In our example, a 4-synapse design was used as a proof-of-concept baseline at various frequencies. Our simulation at 1M Hz indicates a ≈ 91% reduction of energy when using Regenerative Capacitive Synapses vs. standard, nonregenerative ones, which translates into a ≈ 35% drop in overall artificial neuron energy dissipation. The higher the ratio of synapses/soma, the higher the power savings, which is important for building larger and more complex neurons in silico.

An Adiabatic Capacitive Artificial Neuron With RRAM-Based Threshold Detection for Energy-Efficient Neuromorphic Computing

“…The operating principle of the regenerative capacitive synapse has been shown in [29] with basic analysis and energy comparison simulation results versus non-regenerative (nonadiabatic) synapses. In this paper, we substantially extend the analysis, provide mathematical approximations for our fundamental findings, include the comparator/soma in our calculations, perform key parametric analyses indicating how the synaptic loading, temperature and basic process corners affect optimal running frequency and energy dissipation and thus demonstrate a basic, but functionally complete artificial neuron building block circuit that uses capacitive synapses and a RRAM-based threshold detection circuit in depth.…”

“…In this paper, we substantially extend the analysis, provide mathematical approximations for our fundamental findings, include the comparator/soma in our calculations, perform key parametric analyses indicating how the synaptic loading, temperature and basic process corners affect optimal running frequency and energy dissipation and thus demonstrate a basic, but functionally complete artificial neuron building block circuit that uses capacitive synapses and a RRAM-based threshold detection circuit in depth. Additionally, here we use pMOS body biasing to reduce the energy consumption of the capacitive synapses (an upgrade over [29]) and compare the results with the non-adiabatic design.…”