DeepTMA: Predicting Effective Contention Models for Network Calculus using Graph Neural Networks

This repository contains the dataset used for the paper "DeepTMA: Predicting Effective Contention Models for Network Calculus using Graph Neural Networks" publish at the 38th IEEE International Conference on Computer Communications (INFOCOM 2019). We refer to the paper for a full explanation of the methodology used for generating the dataset.

Getting the dataset

The dataset is stored in the git repository using git lfs. Install git lfs on your system first and then clone the repository using:

$ git lfs clone https://github.com/fabgeyer/dataset-infocom2019.git

Reading the dataset

Each file is encoded using Protocol Buffers. The data structure is defined in dataset_infocom2019.proto and can be compiled to various target languages (e.g. Java, Python, Objective-C, and C++) using the protoc command line utility.

Example code in python

The script src/parse_example.py contains an example of how to parse the protobuf files using python. We first compile the .proto file to python:

$ sudo apt install python3-protobuf
$ git clone https://github.com/fabgeyer/dataset-infocom2019
$ cd dataset-infocom2019
$ protoc --python_out=src dataset.proto
$ python src/parse_example.py dataset/dataset.part0.pb.gz

Citation

If you use this dataset for your research, please include the following reference in any resulting publication:

@inproceedings{GeyerBondorf_INFOCOM2019,
	author    = {Geyer, Fabien and Bondorf, Steffen},
	title     = {{DeepTMA}: Predicting Effective Contention Models for Network Calculus using Graph Neural Networks},
	booktitle = {Proceedings of the 38th IEEE International Conference on Computer Communications (INFOCOM)},
	year      = {2019},
	month     = apr,
	address   = {Paris, France},
	doi       = {10.1109/INFOCOM.2019.8737496},
}

License

The data in this repository is licensed under Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0).