GitHub - venkateshsathya_RML22_ Dataset generation code for modulation classification

pdf

School

Concordia University *

*We aren’t endorsed by this school

Course

MAST 233

Subject

Electrical Engineering

Date

Dec 6, 2023

Type

pdf

Pages

3

Report

Uploaded by ChefPower12716

7/7/23, 1:21 PM GitHub - venkateshsathya/RML22: Dataset generation code for modulation classification https://github.com/venkateshsathya/RML22 1/3 venkateshsathya / RML22 Public Dataset generation code for modulation classification 6 stars 1 fork View code RML22 Dataset direct download link . This code is inspired by and reuses part of the code from RADIOML , DeepSig RML22 Dataset generation code details Instructions to run the code. 1. Copy the files under the folder "code" to a local folder "RML22_code". Star Notifications Code Issues Pull requests Actions Projects Security Insights main Go to file vesathya Added tests on RML18 sps = 8.Earlier it was on resampled s on Oct 31, 2022 43
7/7/23, 1:21 PM GitHub - venkateshsathya/RML22: Dataset generation code for modulation classification https://github.com/venkateshsathya/RML22 2/3 2. Copy the Podcast.wav file from this link to "RML22_code". This file is the information source for analog modulation types. 3. This notebook contains cells with code to generate dataset, train on CNN architecure, test and plot results: accuracy versus SNR and confusion matrix. By default it generates RML22 dataset in the first of the notebook. The second cell trains a CNN model with an architecture and training parameters as shown below 5. The third cell tests and plots accuracy versus SNR and a confusion matrix. 6. The five types of datasets that could be generated are clean(no artifacts), with thermal noise over clean dataset (AWGN), with clock effects(SRO/CFO/phase offset) on clean dataset, with fading over clean dataset and the final one RML22 with all artifacts. The default code generates RML22. To generate your own dataset, you need the following GNU radio dependencies. You can also skip installing the dependencies if you do not wish to generate your own dataset and directly download RML22 and train your model on this. Instructions to install GNURadio module and out-of-tree GR-MAPPER module. 1. conda create --name gnuradio 2. conda activate gnuradio 3. conda install -c conda-forge gnuradio=3.8.3 4. conda install -c conda-forge scipy 5. conda install -c conda-forge matplotlib 6. git clone https://github.com/myersw12/gr-mapper.git 7. cd gr-mapper && mkdir build && cd build 8. chmod -R 777 ../../ 9. conda install -c conda-forge gnuradio-build-deps 10. conda activate $CONDA_DEFAULT_ENV 11. conda install -c conda-forge cppunit 12. cmake -G Ninja -DCMAKE_INSTALL_PREFIX=$CONDA_PREFIX - DCMAKE_PREFIX_PATH=$CONDA_PREFIX -DLIB_SUFFIX="" .. 13. cmake --build . README.md
7/7/23, 1:21 PM GitHub - venkateshsathya/RML22: Dataset generation code for modulation classification https://github.com/venkateshsathya/RML22 3/3 List of potential issues. 1. Gr-mapper may or may not work with GNURadio 3.9 See details in link 2. The cmake instructions in this link does not work. Please do not use them. Please follow instructions given above which were taken from CondaInstall . 3. Any issues with cmake: good link to check out is this link "ryanvolz" user on github seems to respond to related queries. 4. If you are getting solving environment error, then do the following "conda config --set channel_priority false". This can help resolve the issue. link 5. If you encounter errors such as gr::vmcircbuf_sysv_shm: shmget (2): No space left on device , you might want to run the program only from command line and not an IDE. We have noticed issues such as low memory allocation when using an IDE that causes this issue. You can also potentially fix the issue by following instrucitons in this link Releases No releases published Packages No packages published Languages Jupyter Notebook 95.2% Python 4.8% 14. cmake --build . --target install
Your preview ends here
Eager to read complete document? Join bartleby learn and gain access to the full version
  • Access to all documents
  • Unlimited textbook solutions
  • 24/7 expert homework help