📄 Read Our Paper 📄

Deep Learning-Based Image Compression for Wireless Communications: Impacts on Reliability, Throughput, and Latency

LIC_TX

Podcast

A discussion on LIC_TX in detail in this podcast episode.

Table of Contents

• Introduction
• Features
• Dependencies
• Installation
• Usage
• Citation
• Contact

Introduction

LIC_TX is a progressive image transmission pipeline designed for dynamic communication channels. It leverages two accessible image autoencoders to ensure efficient and robust image transmission:

1. ProgDTD: A hyperprior-based model (GitHub Repository)
2. VQGAN-LC: A VQGAN-based model (GitHub Repository)

LIC_TX addresses the challenges of transmitting images over dynamic channels by utilizing these advanced autoencoders, providing a seamless and progressive transmission experience.

Features

• Progressive Transmission: Transmit images in a progressive manner, adapting to channel conditions.
• Dynamic Channel Adaptation: Adjusts to varying channel conditions for optimal image quality.
• Residual Quantization: Incorporates residual quantization for efficient data compression.
• Easy Integration: Built upon accessible and widely-used autoencoder models.

Dependencies

LIC_TX requires the following dependencies:

• Software:

  • Python 3.7+
  • pip 23.3.1

• Python Libraries:

  • ldm-lc (Ensure compatibility)
  • ProgDTD
  • compress_AI (Ensure compatibility)
  • PyTorch Lightning (1.7.7)
  • TorchMetrics (0.11.0)
  • faiss-cpu

Installation

Prerequisites

Ensure you have Python 3.7 or higher installed. It's recommended to use a virtual environment.

# Create a virtual environment
python -m venv lic_tx_env

# Activate the virtual environment
# On Windows
lic_tx_env\Scripts\activate
# On Unix or MacOS
source lic_tx_env/bin/activate

Usage

Open LIC_TX.ipynb in the Jupyter interface and run the cells to experiment with the LIC_TX pipeline.

Citation

If you use LIC_TX in your research, please cite our paper:

bibtex

@article{naseri2024deep,
  title={Deep Learning-Based Image Compression for Wireless Communications: Impacts on Reliability, Throughput, and Latency},
  author={Naseri, Mostafa and Ashtari, Pooya and Seif, Mohamed and De Poorter, Eli and Poor, H Vincent and Shahid, Adnan},
  journal={arXiv preprint arXiv:2411.10650},
  year={2024}
}

Contact

For any questions or feedback, please reach out to mostafa.naseri@ugent.be