libmoeprlnc Validator

The tool simulates two communication partners trying to exchange random data with each other using the rlcn library. If both communication partners do not have the same data at the end or unexpected events occur, the program terminates and displays the error on the console. The tool's random nature is aimed at detecting particularly hard-to-find errors.

The tool can also be used for statistical evaluations of the rlcn library.

For more information and documentation, we recommend reading the project report.

Dependencies

This tool requires the installation of the moep80211ncm library, which in turn depends on the libmoep injection library. For more information read the respective README files of the libraries. To install the moep80211ncm use sudo make install as a last step after you followed the library's readme.

Build and Run

To build the tool you can just use make inside the src/moepvalidator which will create a binary called main in the build folder in src/moepvalidator. Run ./build/main --help to see all available arguments or see the section below.

Please note that we have expierenced that on some systems the environment variable LD_LIBRARY_PATH need to be set to /usr/local/lib (or the place where you isntalled libmoep and moep8021ncm) in order for the linker to find the libraries.

Further make options are

• make debug which create build/debug/main without optimization for debugging
• make clean and make clean_debug to delete the respective build files

We have created launch.json and task.json files which define a task that automatically sets the required environment variable, builds the tool and turns on VS Code's debugger.

Program Arguments

    -c, --csv_stats=NAME
    Print statistics to CSV file with the given path.

    -f, --field_size=SIZE
    Set underlying Galois Field size. (0: GF2, 1: GF4, 2: GF16, 3: GF256).

    -g, --gen_size=SIZE
    Set the generation size.

    -i, --nr_iterations=ITER
    Set the amount of test iterations

    -l, --loss_rate=LOSS
    Set probability with which coded data is lost during transmission.

    -m, --mode
    Executes the program in 'pre fill' mode.

    -p, --pkt_size=SIZE
    Set the frame size.

     -s, --seed=ADDR
    Set the seed which is used to generate random test input.

    -v, --verbose
    Produce verbose output.

Python Tool and Jupyter Notebook

We have created a python tool for easy multithreaded execution of the validation tool when one want to test several different configurations. Furthermore, we used a jupyter notebook which uses this tool to create the CSV files of our plots. The notebook also provides code to read out the files and actually created the plots. So given the notebook it should be relatively easy to recreate the plots.

Both files are located in evaluation as eval.ipynb and mass_validation.py. The latter even provides a command line tool. See the inline documentation for further details.

In order to use our python tools one has to install our python dependencies. This can for example be done within a virtual python environment. (Code has only been tested on Python 3.9):

cd evaluation
virualenv --python=python3.9 venv
source venv/bin/activate
pip install -r requirements.txt
# To see the available arguments of the python tool run
python mass_validation.py --help
# To start jupyter notebook run the following command and select the file `eval.ipynb` in your browser
jupyter notebook