ALPaCA

Note we're currently working on deploying ALPaCA as a (Rust) library, libalpaca. If you're interested in the original Python code, please also have a look at the secure drop branch on this repo.

This is the original implementation of the ALPaCA WF defense proposed in ''Website Fingerprinting Defenses at the Application Layer'' (Cherubin, Hayes, Juarez. PETS'17).

The defense can operate in a probabilistic (P-ALPaCA) or a deterministic (D-ALPaCA) mode.

Note: this is a research prototype. Please, do not use this code to defend your website as yet, as it needs peer-reviewing. We will remove this disclaimer when appropriate.

Requirements

Coming soon.

P-ALPaCa

Morphing a page $PAGE. The morphed page is put into directory $DST. The following assumes you have generated the KDE distributions, and put them into directory $DISTD. You can find the distributions we used for experiments in data/distributions/.

export PAGE=                  # .html file to morph.
export DST=                   # destination directory for the morphed page.
export DISTD=                 # Folder containing the KDE distributions.

python ssd.py --page $PAGE --dst $DST distribution --distribution-type kde --count-dist $DISTD/counts.kde --html-dist $DISTD/html.kde --objects-dist $DISTD/objects.kde

D-ALPaCa

Morphing a page $PAGE. The morphed page is put into directory $DST.

export L=       # Parameter "lambda".
export S=       # Parameter "sigma"
export MAXS=    # Parameter "max_s"

python ssd.py --page $PAGE --dst $DST deterministic --L $L --S $S --maxs $MAXS

Generating custom distributions for P-ALPaCA

Coming soon.