We here make available the code and datasets for our paper “When the Signal is in the Noise: Exploiting Diffix’s Sticky Noise”. The paper has been accepted at USENIX Security 2019 and can be downloaded here. The full version of the paper is available on ArXiv.

Citation

If you use any of the materials, please cite the paper as:

Gadotti, A., Houssiau, F., Rocher, L., Livshits, B., de Montjoye, Y.-A. When the Signal is in the Noise: Exploiting Diffix’s Sticky Noise. In Proceedings of the 28th USENIX Security Symposium, 2019.

Code

The code for the experiments can be downloaded here. The archive contains the code for the attacks and the Jupyter notebooks used to perform the experiments that appear in the paper. For intellectual property reasons, the archive does not contain the code of our (partial) implementation of the Diffix mechanism. We however added a README that describes how the mechanism has been implemented.

Datasets

We performed the attacks on four datasets. Three datasets are publicly available and can be downloaded from the UCI Machine Learning Repository website:

• ADULT
• CENSUS
• CREDIT

Additionally, we generated a synthetic CDR dataset containing fake trajectories for 2M users which we will make available upon publication of the related paper.