Energy-Efficient Near-Sensor Event Detector Based on Multilevel Ga2O3 RRAM

Mehrdad Morsali, Sepehr Tabrizchi, Ravi Teja Velpula, Mano Bala Sankar Muthu, Hieu Pham Trung Nguyen, Mohsen Imani, Arman Roohi, Shaahin Angizi

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

In this paper, a cost-effective Near-Sensor Processing (NSP) platform is developed based on an experimentally-measured TiTiN/Ga2O3/Ti/Pt Resistive Random Access Memory (RRAM) device that facilitates event detection for edge vision sensors without the requirement for power-intensive Analog-to-Digital Converters (ADCs). The platform is supported with a hardware-friendly background comparison technique providing adjustable precision that allows for a dynamic balance between accuracy and efficiency at runtime. Our device-to-architecture simulation results demonstrate that the proposed platform achieves on average 66% and 63% energy saving over STT-MRAM and SOT-MRAM counterparts due to utilizing the ADC-less method.

Original languageEnglish (US)
Title of host publication2024 IEEE Computer Society Annual Symposium on VLSI
Subtitle of host publicationEmerging VLSI Technologies and Architectures, ISVLSI 2024
EditorsHimanshu Thapliyal, Jurgen Becker
PublisherIEEE Computer Society
Pages331-336
Number of pages6
ISBN (Electronic)9798350354119
DOIs
StatePublished - 2024
Event2024 IEEE Computer Society Annual Symposium on VLSI, ISVLSI 2024 - Knoxville, United States
Duration: Jul 1 2024Jul 3 2024

Publication series

NameProceedings of IEEE Computer Society Annual Symposium on VLSI, ISVLSI
ISSN (Print)2159-3469
ISSN (Electronic)2159-3477

Conference

Conference2024 IEEE Computer Society Annual Symposium on VLSI, ISVLSI 2024
Country/TerritoryUnited States
CityKnoxville
Period7/1/247/3/24

All Science Journal Classification (ASJC) codes

  • Hardware and Architecture
  • Control and Systems Engineering
  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'Energy-Efficient Near-Sensor Event Detector Based on Multilevel Ga2O3 RRAM'. Together they form a unique fingerprint.

Cite this