A Generative Self-supervised Framework for Cognitive Radio Leveraging Time-Frequency Features and Attention-based Fusion
This repository provides the implementation of the generative self-supervised framework described in the following paper:
Chen, Shuai; Feng, Zhixi; Yang, Shuyuan; Ma, Yue; Liu, Jun; Qi, Zhuoyue
A Generative Self-supervised Framework for Cognitive Radio Leveraging Time-Frequency Features and Attention-based Fusion.
IEEE Transactions on Wireless Communications, 2024.
DOI: 10.1109/TWC.2024.3513980
Ensure you have Python installed (version >= 3.8 is recommended). Install the required dependencies using the provided requirements.txt
file:
Follow the dataset processing instructions provided in the paper. The datasets used in the experiments are:
- RadioML2016.10a
- RadioML2016.10b
- RML2016.04c
- ADS-B short dataset
Execute the pretrain.py script to perform pretraining:
python pretrain.py
This script will:
- Load the datasets from the data/ directory.
- Perform self-supervised pretraining using the settings described in the paper.
If you find this repository useful in your work, please consider citing the following paper:
@ARTICLE{10804099,
author={Chen, Shuai and Feng, Zhixi and Yang, Shuyuan and Ma, Yue and Liu, Jun and Qi, Zhuoyue},
journal={IEEE Transactions on Wireless Communications},
title={A Generative Self-supervised Framework for Cognitive Radio Leveraging Time-Frequency Features and Attention-based Fusion},
year={2024},
volume={},
number={},
pages={1-1},
keywords={Feature extraction;Time-frequency analysis;Data mining;Spectrogram;Transformers;Modulation;Radio communication;Noise reduction;Cognitive radio;Time-domain analysis;Generative framework;self-supervised learning (SSL);cognitive radio technology (CRT)},
doi={10.1109/TWC.2024.3513980}
}