Smartphone identification via passive traffic fingerprinting

This repository contains the reference code for the paper ''Smartphone Identification via Passive Traffic Fingerprinting: a Sequence-to-Sequence Learning Approach'' DOI: 10.1109/MNET.001.1900101.

If you find the project useful and you use this code, please cite our paper:

@article{Meneghello2020Network,
	author={Francesca Meneghello and Michele Rossi and Nicola Bui},
	title={Smartphone Identification via Passive Traffic Fingerprinting: a Sequence-to-Sequence Learning Approach},
	journal={IEEE Network},
	volume={34},
	number={2},
	pages={112--120},
	year={2020}
}

How to use

Clone the repository and enter the folder with the python code:

cd <your_path>
git clone https://github.com/francescamen/smartphone_identification
cd code

Download the input data from http://researchdata.cab.unipd.it/id/eprint/292, unzip and put them into the input_files folder.

Train and test the framework

To create the smartphone fingerprints and uses them to correctly associate unknown traffic traces to the user labels execute the following commands:

python data_loading_preprocessing.py 
python train_denoising_autoencoder.py <hidden_neurons> <layers> <epochs_RNN>
python run_encoder.py <hidden_neurons> <layers>
python words_clustering.py <num_clusters>
python users_identification.py <num_clusters> <epochs_CNN>

Visualize the results

In the code folder you can find other utilities functions. To visualize the performance of the autoencoder run the following command:

python plot_autoencoder.py <hidden_neurons> <layers>

The confusion matrix can be computed and plotted throught the command:

python confusion_matrix_analysis.py 

The users similarity assessment is performed by executing the following commands:

python users_disambiguation.py <num_clusters> <epochs_CNN>
python Bhattacharyya_distance.py <num_clusters> 

Parameters

The results on the article are obtained with the following parameters: <hidden_neurons>=32 <layers>=2 <epochs_RNN>=100 <num_clusters>=50 <epochs_CNN>=100.

Authors

Francesca Meneghello, Michele Rossi, Nicola Bui

Contact

[email protected] github.com/francescamen