Skip to main navigation Skip to search Skip to main content

Deep Joint Source Channel Coding via Attention for Wireless Image Transmission

  • Haoze Chang
  • , Lin Ma*
  • , Xuedong Wang
  • *Corresponding author for this work
  • School of Electronics and Information Engineering, Harbin Institute of Technology

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

In digital communication, efficiently transmitting image and video data through constrained channels remains challenging nowadays. Traditional methods using separate source and channel coding often fail in dynamic environments. In this paper, we introduce a novel deep learning based (DL) attention joint source channel coding (AttenJSCC) approach, which enhances robustness and efficiency in wireless image transmissions. By integrating source and channel coding into a unified framework and incorporating our Enhanced Attention Feature (EAF) modules and the ECA attention mechanism, our method outperforms some of the existing JSCC techniques, especially in low SNR conditions. Our framework not only overcomes the limitations of current technologies but also reduces the storage and computational needs on edge devices, facilitating more efficient real time communication.

Original languageEnglish
Title of host publicationWireless and Satellite Systems - 14th EAI International Conference, WiSATS 2024, Proceedings
EditorsHsiao-Hwa Chen, Weixiao Meng
PublisherSpringer Science and Business Media Deutschland GmbH
Pages312-323
Number of pages12
ISBN (Print)9783031861956
DOIs
StatePublished - 2025
Externally publishedYes
Event14th EAI International Conference on Wireless and Satellite Systems, WiSATS 2024 - Harbin, China
Duration: 23 Aug 202425 Aug 2024

Publication series

NameLecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICST
Volume605 LNICST
ISSN (Print)1867-8211
ISSN (Electronic)1867-822X

Conference

Conference14th EAI International Conference on Wireless and Satellite Systems, WiSATS 2024
Country/TerritoryChina
CityHarbin
Period23/08/2425/08/24

Keywords

  • Attention Mechanisms
  • Deep Learning
  • JSCC
  • Wireless Image Transmission

Fingerprint

Dive into the research topics of 'Deep Joint Source Channel Coding via Attention for Wireless Image Transmission'. Together they form a unique fingerprint.

Cite this