Fork of adeeb10abbas/RadioML_2016, which is a fork of radioML/dataset
Remove dependencies on gr-mapper and gr-mediatools
- python
- gnuradio
- numpy
- scipy
conda create -n gnuradio
conda activate gnuradio
conda config --env --add channels conda-forge
conda config --env --set channel_priority strict
conda install numpy scipy gnuradio=3.8
First, create an output folder
mkdir datasets
There are two constants in generate_RML2016.10a.py
:
apply_channel
- whether or not to model the dynamic transmission channelthermal_noise
- whether or not to model AWGN at the receiver Set these parameters to your preference.
Then run:
python generate_RML2016.10a.py
- We are using wavfile source instead of gr-mediatools.
- We are using built in modulations instead of gr-mapper
- Chnage analog source from serial podcast (which has a lot of silence) to public domain recording of The Star Spangled Banner (which was editted to remove silence at the beginning of the recording).
- Add option to turn off AWGN