Estimating Confidentiality of Encrypted Search Systems: A Probabilistic Framework

This repository contains research on measuring the confidentiality of encrypted search systems against an adversary that observes hidden queries to infer plaintext queries.

Abstract

Encrypted search systems enable privacy-preserving queries over confidential data stored on untrusted servers. However, the confidentiality of such systems may be compromised through frequency analysis attacks. We develop a probabilistic framework for quantifying the confidentiality of encrypted search systems based on the sampling distribution of a confidentiality statistic. Using the Bootstrap method with normal approximation, we efficiently estimate the risk of confidentiality breach, enabling proactive countermeasures. Our analysis shows how entropy of the query distribution relates to the adversary's expected accuracy, providing theoretical grounding for resilience engineering approaches to encrypted search security.

Repository Structure

.
├── paper/              # LaTeX paper source
│   ├── main.tex        # Main document
│   ├── refs.bib        # Bibliography
│   ├── Makefile        # Build automation
│   ├── sections/       # LaTeX section files
│   ├── images/         # Figures
│   └── data/           # Experimental results
├── src/                # C++ simulation code
├── CITATION.cff        # Citation metadata
└── LICENSE             # CC-BY 4.0

Building the Paper

cd paper
make          # Build PDF with bibliography
make quick    # Quick build (no bibliography)
make clean    # Remove auxiliary files

Citation

If you use this work, please cite:

@article{towell2024estimating,
  title={Estimating Confidentiality of Encrypted Search Systems: A Probabilistic Framework},
  author={Towell, Alex},
  year={2024}
}

License

This work is licensed under the Creative Commons Attribution 4.0 International License.