中文

矢量量化语义通信系统

计算机视觉与模式识别 2023-04-14 v2 信号处理

摘要

尽管模拟语义通信系统已在文献中受到相当关注,关于数字语义通信系统的工作仍较少。本文中,我们开发了一种用于图像传输的深度学习(DL)赋能矢量量化(VQ)语义通信系统,名为VQ-DeepSC。具体而言,我们提出基于卷积神经网络(CNN)的收发机以提取图像的多尺度语义特征,并引入多尺度语义嵌入空间来执行语义特征量化,使数据与数字通信系统兼容。此外,我们通过引入PatchGAN判别器采用对抗训练以提升接收图像的质量。实验结果表明,所提VQ-DeepSC在数字通信系统中比BPG更具鲁棒性,且具有与DeepJSCC方法相当的MS-SSIM性能。

关键词

引用

@article{arxiv.2209.11519,
  title  = {Vector Quantized Semantic Communication System},
  author = {Qifan Fu and Huiqiang Xie and Zhijin Qin and Gregory Slabaugh and Xiaoming Tao},
  journal= {arXiv preprint arXiv:2209.11519},
  year   = {2023}
}

备注

This five pages article has been accepted for publication in IEEE Wireless Communications Letters. This is the author's version which has not been fully edited and content may change prior to final publication. Citation information: DOI 10.1109/LWC.2023.3255221