Eciton: Very Low-Power LSTM Neural Network Accelerator for Predictive Maintenance at the Edge

This paper presents Eciton, a very low-power LSTM neural network accelerator for low-power edge sensor nodes, demonstrating real-time processing on predictive maintenance applications with a power consumption of 17 mW under load. Eciton reduces memory and chip resource requirements via 8-bit quantiz...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:International Conference on Field-programmable Logic and Applications S. 1 - 8
Hauptverfasser: Chen, Jeffrey, Hong, Sehwan, He, Warrick, Moon, Jinyeong, Jun, Sang-Woo
Format: Tagungsbericht
Sprache:Englisch
Veröffentlicht: IEEE 01.08.2021
Schlagworte:
CPS
edge
iCE40
IoT
Lattices
low power
LSTM
Memory management
Neural networks
Power demand
Predictive Maintenance
Quantization (signal)
Wireless communication
Wireless sensor networks
ISSN:1946-1488
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Abstract This paper presents Eciton, a very low-power LSTM neural network accelerator for low-power edge sensor nodes, demonstrating real-time processing on predictive maintenance applications with a power consumption of 17 mW under load. Eciton reduces memory and chip resource requirements via 8-bit quantization and hard sigmoid activation, allowing the accelerator as well as the LSTM model parameters to fit in a lowcost, low-power Lattice iCE40 UP5K FPGA. Eciton demonstrates real-time processing at a very low power consumption with minimal loss of accuracy on two predictive maintenance scenarios with differing characteristics, while achieving competitive power efficiency against the state-of-the-art of similar scale. We also show that the addition of this accelerator actually reduces the power budget of the sensor node by reducing power-hungry wireless transmission.The resulting power budget of the sensor node is small enough to be powered by a power harvester, potentially allowing it to run indefinitely without a battery or periodic maintenance.
AbstractList This paper presents Eciton, a very low-power LSTM neural network accelerator for low-power edge sensor nodes, demonstrating real-time processing on predictive maintenance applications with a power consumption of 17 mW under load. Eciton reduces memory and chip resource requirements via 8-bit quantization and hard sigmoid activation, allowing the accelerator as well as the LSTM model parameters to fit in a lowcost, low-power Lattice iCE40 UP5K FPGA. Eciton demonstrates real-time processing at a very low power consumption with minimal loss of accuracy on two predictive maintenance scenarios with differing characteristics, while achieving competitive power efficiency against the state-of-the-art of similar scale. We also show that the addition of this accelerator actually reduces the power budget of the sensor node by reducing power-hungry wireless transmission.The resulting power budget of the sensor node is small enough to be powered by a power harvester, potentially allowing it to run indefinitely without a battery or periodic maintenance.
Author Chen, Jeffrey
He, Warrick
Moon, Jinyeong
Hong, Sehwan
Jun, Sang-Woo
Author_xml – sequence: 1
  givenname: Jeffrey
  surname: Chen
  fullname: Chen, Jeffrey
  email: jeffrc2@uci.edu
  organization: University of California,Department of Computer Science,Irvine
– sequence: 2
  givenname: Sehwan
  surname: Hong
  fullname: Hong, Sehwan
  email: sehwanh@uci.edu
  organization: University of California,Department of Computer Science,Irvine
– sequence: 3
  givenname: Warrick
  surname: He
  fullname: He, Warrick
  email: warrickhe@gmail.com
  organization: Diamond Bar High School
– sequence: 4
  givenname: Jinyeong
  surname: Moon
  fullname: Moon, Jinyeong
  email: j.moon@fsu.edu
  organization: Florida State University,Department of Electrical and Computer Engineering
– sequence: 5
  givenname: Sang-Woo
  surname: Jun
  fullname: Jun, Sang-Woo
  email: swjun@uci.edu
  organization: University of California,Department of Computer Science,Irvine
BookMark eNotjt9KwzAYxaMouM09wW7yAp358qdpvBujU6HTgtM7GTH9qtHaShote3sr-oPDuTnncKbkpO1aJGQBbAnAzMWmLJTQJltyxmHJRswRmUKaKim0MukxmYCRaQIyy87IvO_ffjNK6kylE_KUOx-79pI-YjjQohuSshsw0OJ-t6W3-BVsM1ocuvBOV85hg8HGLtB6VBmw8i76b6Rb69uIrW0dUhtpfEWaVy94Tk5r2_Q4__cZedjku_V1Utxd3axXReK5zGIia8klq8EKMKzimElnNSAzXKlUVsoo_sxBKMm0HL-jrZzhGmoLtRFag5iRxd-uR8T9Z_AfNhz2ZmyLkR-JZ1Oy
CODEN IEEPAD
ContentType Conference Proceeding
DBID 6IE
6IL
CBEJK
RIE
RIL
DOI 10.1109/FPL53798.2021.00009
DatabaseName IEEE Electronic Library (IEL) Conference Proceedings
IEEE Xplore POP ALL
IEEE Xplore All Conference Proceedings
IEEE Electronic Library (IEL)
IEEE Proceedings Order Plans (POP All) 1998-Present
DatabaseTitleList
Database_xml – sequence: 1
  dbid: RIE
  name: IEEE Electronic Library (IEL)
  url: https://ieeexplore.ieee.org/
  sourceTypes: Publisher
DeliveryMethod fulltext_linktorsrc
EISBN 1665437596
9781665437592
EISSN 1946-1488
EndPage 8
ExternalDocumentID 9556333
Genre orig-research
GroupedDBID 6IE
6IF
6IL
6IN
AAWTH
ABLEC
ADZIZ
ALMA_UNASSIGNED_HOLDINGS
BEFXN
BFFAM
BGNUA
BKEBE
BPEOZ
CBEJK
CHZPO
IEGSK
OCL
RIE
RIL
ID FETCH-LOGICAL-i248t-4f4240f1a3190d2e84ca71e0925564d5952b21354074000eadc9271fa1f937713
IEDL.DBID RIE
ISICitedReferencesCount 66
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000728589800001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
IngestDate Wed Aug 27 02:25:25 EDT 2025
IsDoiOpenAccess false
IsOpenAccess true
IsPeerReviewed false
IsScholarly false
Language English
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-i248t-4f4240f1a3190d2e84ca71e0925564d5952b21354074000eadc9271fa1f937713
OpenAccessLink https://escholarship.org/uc/item/8p35m8pz
PageCount 8
ParticipantIDs ieee_primary_9556333
PublicationCentury 2000
PublicationDate 2021-Aug.
PublicationDateYYYYMMDD 2021-08-01
PublicationDate_xml – month: 08
  year: 2021
  text: 2021-Aug.
PublicationDecade 2020
PublicationTitle International Conference on Field-programmable Logic and Applications
PublicationTitleAbbrev FPL
PublicationYear 2021
Publisher IEEE
Publisher_xml – name: IEEE
SSID ssj0000547856
Score 2.072466
Snippet This paper presents Eciton, a very low-power LSTM neural network accelerator for low-power edge sensor nodes, demonstrating real-time processing on predictive...
SourceID ieee
SourceType Publisher
StartPage 1
SubjectTerms CPS
edge
iCE40
IoT
Lattices
low power
LSTM
Memory management
Neural networks
Power demand
Predictive Maintenance
Quantization (signal)
Wireless communication
Wireless sensor networks
Title Eciton: Very Low-Power LSTM Neural Network Accelerator for Predictive Maintenance at the Edge
URI https://ieeexplore.ieee.org/document/9556333
WOSCitedRecordID wos000728589800001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3NS8MwFA9zePCksonf5ODRuqZNmsSbyIaHbhScYxcZafIqA-lkdor_vXnZmAhePCXpIYWXvP5e38fvEXKVaqWEU1i6VdqIMycjI0UVAWRGSa4qHbpETHI5GqnpVBctcr2thQGAkHwGNzgNsXy3sCt0lfU0slml6Q7ZkVKua7W2_pQYialEtiEWYrHuDYpcpFJj_lbCAk-h_tVCJSDIYP9_7z4g3Z9SPFpsQeaQtKDukOe-9YpY39IJLL9ovviMCmx2RvPH8ZAi3YZ59UPI76Z31npkCcF06g1UvxmGZvAjR4cGySKQcQOoaag3BWnfvUCXPA364_uHaNMnIZonXDURr7jH5YoZr06xS0BxaySDWCO9GHdCi6RMGDp4pNfY2N8dqxPJKsMqb5z4v9Qj0q4XNRwTWupSqNIBF-CxuyxNBtIvs1iDy1yiT0gHRTN7W1NhzDZSOf378RnZQ9mv8-XOSbtZruCC7NqPZv6-vAzn9w0sEZr6
linkProvider IEEE
linkToHtml http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3NS8MwFA9zCnpS2cRvc_BoXZMmTeJNZGNiNwbOsYuMNHmVgXQyO8X_3iQbE8GLpyQ9pJCX19_r-_g9hC4TJSW30pdu5SZixIpIC15EAKmWgslChS4Ro0z0-3I8VoMaulrXwgBASD6Daz8NsXw7MwvvKmspz2aVJBtokzNGybJaa-1RiT01FU9X1EIkVq3OIOOJUD6Di5LAVKh-NVEJGNLZ_d_b91DzpxgPD9Yws49qUDbQc9s4VSxv8AjmXzibfUYD3-4MZ4_DHvaEG_rVDSHDG98a47AlhNOxM1HdZj444z9zuKc9XYTn3ACsK-yMQdy2L9BET5328K4brTolRFPKZBWxgjlkLoh2ChVbCpIZLQjEyhOMMcsVpzkl3sUjnM7G7vYYRQUpNCmceeL-Uw9QvZyVcIhwrnIucwuMg0PvPNcpCLdMYwU2tVQdoYY_msnbkgxjsjqV478fX6Dt7rCXTbL7_sMJ2vFyWGbPnaJ6NV_AGdoyH9X0fX4eZPkNbLueQQ
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=proceeding&rft.title=International+Conference+on+Field-programmable+Logic+and+Applications&rft.atitle=Eciton%3A+Very+Low-Power+LSTM+Neural+Network+Accelerator+for+Predictive+Maintenance+at+the+Edge&rft.au=Chen%2C+Jeffrey&rft.au=Hong%2C+Sehwan&rft.au=He%2C+Warrick&rft.au=Moon%2C+Jinyeong&rft.date=2021-08-01&rft.pub=IEEE&rft.eissn=1946-1488&rft.spage=1&rft.epage=8&rft_id=info:doi/10.1109%2FFPL53798.2021.00009&rft.externalDocID=9556333