KANSC

Zifan Zhu, Computer Science of Western University

KAN integration of Deep Learning Enabled Semantic Communication Systems

DeepSC author:

Huiqiang Xie, Zhijin Qin, Geoffrey Ye Li, and Biing-Hwang Juang

Requirements

• Refer to readme of this repository for setting up Pykan environment

• See the requirements.txt for the required python packages for original DeepSC and run pip install -r requirements.txt to install them.

Preprocess

mkdir data
wget http://www.statmt.org/europarl/v7/europarl.tgz
tar zxvf europarl.tgz
python preprocess_text.py

Train

• For training KANSC model

python main_kan.py

• For training DeepSC model

python main.py

Evaluation

Check kan_sc_performance.ipynb file for evaluation of KANSC model Check performance.ipynb file for evaluation of DeepSC model

Notes

• The Integrated version only focuses on evaluating the BLEU score performance vs. different SNR over AWGN and Rayleigh Fading Channel

• The efficient_kan implementation of KAN layer is from this repository

Bibtex

@article{xie2021deep,
  author={H. {Xie} and Z. {Qin} and G. Y. {Li} and B. -H. {Juang}},
  journal={IEEE Transactions on Signal Processing},
  title={Deep Learning Enabled Semantic Communication Systems},
  year={2021},
  volume={Early Access}}