ECNet: a fast, accurate, and lightweight edge-cloud network system based on cascading structure

Libo Hu, Tao Wang, Hiroshi Watanabe, Shohei Enomoto, Xu Shi, Akira Sakamoto, Takeharu Eda

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

1 Citation (Scopus)

Abstract

The pervasiveness of 'Internet-of-Things' in daily life has led to a recent surge in fog computing, encompassing a collaboration of cloud computing and edge intelligence. As a significant field of IoT, real-time detection and classification have a huge demand. Due to the insufficiency of computing power in mobile devices and the increment of network bandwidth, combination of edge devices and cloud servers would be an accessible orientation for real-time tasks. In this work, we present ECNet - an edge-cloud network system dealing with the balance between performance and time cost. We propose to transmit the output ferefature map of an exit point to the cloud with offload controller and quantizer deployed to minimize the transmission cost. ECNet is tested to reach a balance between processing time and accuracy performance with reducing transmission cost down to 25%. We also consider implementing an integrated feature map encoder to further reduce the bandwidth demand and meanwhile minimize the loss of accuracy. Additional achievements could be expected in our future work.

Original languageEnglish
Title of host publication2020 IEEE 9th Global Conference on Consumer Electronics, GCCE 2020
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages186-189
Number of pages4
ISBN (Electronic)9781728198026
DOIs
Publication statusPublished - 2020 Oct 13
Event9th IEEE Global Conference on Consumer Electronics, GCCE 2020 - Kobe, Japan
Duration: 2020 Oct 132020 Oct 16

Publication series

Name2020 IEEE 9th Global Conference on Consumer Electronics, GCCE 2020

Conference

Conference9th IEEE Global Conference on Consumer Electronics, GCCE 2020
Country/TerritoryJapan
CityKobe
Period20/10/1320/10/16

Keywords

  • ECNet
  • Edge-cloud system
  • classification
  • encoding
  • model cascading

ASJC Scopus subject areas

  • Signal Processing
  • Electrical and Electronic Engineering
  • Media Technology
  • Instrumentation
  • Computer Networks and Communications
  • Computer Vision and Pattern Recognition

Fingerprint

Dive into the research topics of 'ECNet: a fast, accurate, and lightweight edge-cloud network system based on cascading structure'. Together they form a unique fingerprint.

Cite this