drexelwireless/radioml

Code for "Radio Modulation Classification Using Deep Residual Neural Networks" paper

modrec

Run the following commands after downloading the datasets from deepsig.io-

pip install -e .
make

For training -

Example --

python3 bin/train.py data/2016.10a.h5 --train --model resnet18-outer --batch-size 512 

To Cite

  author={Abbas, Adeeb and Pano, Vasil and Mainland, Geoffrey and Dandekar, Kapil},
  booktitle={MILCOM 2022 - 2022 IEEE Military Communications Conference (MILCOM)}, 
  title={Radio Modulation Classification Using Deep Residual Neural Networks}, 
  year={2022},
  volume={},
  number={},
  pages={311-317},
  doi={10.1109/MILCOM55135.2022.10017640}}

This repo also has other supporting tools like data loaders/saving it in hdf5 etc that were used during the work - https://ieeexplore.ieee.org/abstract/document/10017640/

There are no active maintainers of this project. For any queries/concerns, email - [email protected]

radioml's Issues

Numba/PyCuda use for the rgb encoded dataset generation

@BillJr99 @mainland Do you think we could use something like this to make the dataset generation even faster? Especially because since I have a decent GPU(2080ti)

I tried doing it but I am struggling with a few things. Let me know what you think.

Adeeb Abbas