This repository contains the code used to simulate the attacks presented in our CODASPY '24 Paper, "Exploiting Update Leakage in Searchable Encryption"
First, download the code into the environment that you would like to run the attacks.
Unzip datasets.zip directly into your working directory. Your working directory should be flat after this step.
Next, install NumPy and SciPy into your virtual environment.
Run the experiments using the command below. This will take a long time - we recommend running the experiments overnight.
python3 automate_experiments.py
The experiments run in the order that they are listed in the exp_settings.csv file.
As experiments are completed, you will see intermediate result files appear in the experiment_results/ directory.
If for any reason the experiments stop (crash, want to stop and continue later, etc.), place a # in front of all lines for experiments
in exp_settings.csv that have already been ran (or simply delete them).
Before the results can be graphed, we have an intermediate processing step to combine the results for each type of experiment into one file.
Run the following command:
python3 output_to_graph_format.py
You should see 10 files appear in the results_for_latex directory.
Open main.tex in your favorite LaTeX editor, such as OverLeaf. Drop the results_for_latex/ folder into the environment (i.e., do not put the contents of results_for_latex/ into the same level as main.tex, the working directory should not be flat).
Compile main.tex and view the results of the graphs. We use constant values in 4 places in the figures; please Ctrl-F for "Constant" and verify that each of the values match the results from your eta_mu.txt. If not, replace those constants with the values you have.
We use Linux-style paths which may break on Windows. We recommend running the experiment code in a fairly strong Linux environment. Multithreading could speed up the experiments, but we do not have plans to implement it.
We have wrapped the logic used to generate our paper's experimental results to automate our previously-tedious workflow; the result is the code in this repo. The intent and make it user-friendly for readers to verify our results. We have done some testing to make sure that the code in this repository works, but we expect this process to have introduced some bugs. If you run into bugs or crashes, please don't spend your time trying to fix them - report them to jacobshalt(at)vt(dot)edu and we will do our best to fix them promptly.
This code may be used for academic research only. All other uses are prohibited.