Energy-Efficient Ferroelectric Field-Effect Transistor-Based Oscillators for Neuromorphic System Design

Neuromorphic or bioinspired computational platforms, as an alternative for von-Neumann structures, have benefitted from the excellent features of emerging technologies in order to emulate the behavior of the biological brain in an accurate and energy-efficient way. Integrability with CMOS technology...

Full description

Saved in:
Bibliographic Details
Published in:IEEE journal on exploratory solid-state computational devices and circuits 2020-12, Vol.6 (2), p.122-129
Main Authors: Eslahi, Hossein, Hamilton, Tara J., Khandelwal, Sourabh
Format: Article
Language:English
Subjects:
Biomimetics
CMOS
CMOS technology
coupled oscillator
Edge detection
Energy efficiency
FeFETs
ferroelectric field-effect transistor (FEFET) device
Ferroelectric materials
Ferroelectricity
Field effect transistors
Frequency modulation
Image processing
Integrated circuit modeling
Neural networks
neuromorphic computing
Neuromorphic engineering
Neurons
New technology
Oscillators
Power consumption
Semiconductor devices
Systems design
Transistors
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Neuromorphic or bioinspired computational platforms, as an alternative for von-Neumann structures, have benefitted from the excellent features of emerging technologies in order to emulate the behavior of the biological brain in an accurate and energy-efficient way. Integrability with CMOS technology and low power consumption make ferroelectric field-effect transistor (FEFET) an attractive candidate to perform such paradigms, particularly for image processing. In this article, we use the FEFET device to make energy-efficient oscillatory neurons as the main parts of neural networks for image processing applications, especially for edge detection. Based on our simulation results, we estimated a significant energy efficiency compared with other technologies, which shows roughly 5-120\times reduction, depending on the design.
ISSN:2329-9231
2329-9231
DOI:10.1109/JXCDC.2020.3027541