An-Efficient-RFF-Extraction-Method

This is a PyTorch/GPU implementation of the paper An Efficient RFF Extraction Method Using Asymmetric Masked Auto-Encoder. If using relevant content, please cite this paper:

@INPROCEEDINGS{10460605,
  author={Yao, Zhisheng and Fu, Xue and Wang, Shufei and Wang, Yu and Gui, Guan and Mao, Shiwen},
  booktitle={2023 28th Asia Pacific Conference on Communications (APCC)}, 
  title={An Efficient RFF Extraction Method Using Asymmetric Masked Auto-Encoder}, 
  year={2023},
  volume={},
  number={},
  pages={364-368},
  keywords={Convolutional codes;Wireless communication;Training;Convolution;Fingerprint recognition;Feature extraction;Transceivers;Radio frequency fingerprint (RFF);unsupervised learning;asymmetric masked auto-encoder (AMAE)},
  doi={10.1109/APCC60132.2023.10460605}}

• Attention, you need to manually create three folders and name them as 'model_weight' and 'test_result'.
• In addition, the dataset link used in this demo is https://ieee-dataport.org/open-access/lorarffidataset.

Catalog

• Training code
• Few-shot training code
• AWGN training code
• Visualization code

Training and Visualization

• Start training by running the train_FS-AMAE.py or train_FS-AMAE.py file.
• After the training is completed, the Visualization.py file can be run to visualize the features of the trained model, and the trained model can be evaluated using unsupervised clustering indicators.