Repository containing the code of FP-Scanner's paper.

Create virtual environment and install dependencies

Run the command below to create a virtual environment.

virtualenv --python=/usr/bin/python3 myvenv

Then activate the virtual environment.

. myvenv/bin/activate

Finally, install the dependencies.

pip install -r requirements.txt

Import database

Our prototype relies on MongoDB to store the fingerprints, thus you need to ensure that MongoDB is running.

Run the command below to import the database.

mongoimport --db usenix18 --collection fingerprint --file fingerprints.json 

Scan fingerprints

First, we analyze the fingerprints present in the database.

python main.py

It logs information on the screen about the results of the analysis and generates two csv result files in the results folder:

• res_prediction.csv that contains information on the tests that passed or not for each fingerprint;
• res_real_values.csv that contains information on the OS and browser predicted for each fingerprint.

Then we analyse these files to obtain the accuracy of FP-Scanner, FingerprintJS2 and Augur.

python main.py analyse

Benchmark

To run the benchmark that measures the execution time of the scanner, run the command below.

python main.py bench

It generates three files bench_situation1.csv, bench_situation2.csv and bench_situation3.csv that corresponds to the three cases presented in the article:

1. Run all scanner tests even when an inconsistency is detected;
2. Stop running tests when an inconsistency is detected;
3. Run only analysis of the pixels.

These files contain a single column called elapsed_time, which represent the execution time needed to run the set of tests.