MARS is a system for multi-key searchable encryption that allows multiple data owners to store and share encrypted multimaps with clients while maintaining security against arbitrary collusion. It uses unkeyed PIR augmented with flexible-sharing mechanisms for efficient searching without data replication.

⚠️ Research Use Only

This implementation was built specifically to produce data for our research paper. While it's a decent implementation for other researchers to use in their academic work, it is not recommended for production use. The code has not been audited for security vulnerabilities and may not implement all necessary safeguards for real-world applications.

If you plan to do searchable encryption in a production environment, you should seek or create a properly audited implementation. If you create a production-ready version of this code and can verify its security, please let me know and I'll be happy to link to your repository.

• Unkeyed PIR with flexible-sharing mechanisms.
• Support for large-scale multimaps (tested up to 24M entries).
• Query time of ~1s for keywords matching 100 documents. Scales linearly with the number of shards.
• Storage overhead factor of ~6x.

1. Install the latest stable version of Rust and Cargo:

   curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh   # On Unix/macOS
   # OR visit https://rustup.rs/ for Windows installation

1. Create and activate a virtual environment:

   python3 -m venv env
   source env/bin/activate  # On Unix/macOS
   env\Scripts\activate     # On Windows

2. Install required packages:

   pip install nltk numpy tqdm

3. Download required NLTK data:

   python3 -c "import nltk; nltk.download('punkt'); nltk.download('stopwords'); nltk.download('words')"

1. Generate test data:

   cd data
   python generate_power.py  # For power-law distributed synthetic data
   # OR
   python parse_enron.py    # For processing Enron dataset

2. Build and run:

   cargo build --release
   ./target/release/mkse --data-dir data

The project includes several test scenarios:

# Run all tests
DATA_DIR=<data> USE_EXISTING=false make all

# Run sharding tests
DATA_DIR=<data> USE_EXISTING=true make sharding

# Run subset tests
DATA_DIR=<data> USE_EXISTING=true make subsets

# Run sharding tests with keywords linked to $|$Docs_w$|$ documents
DATA_DIR=<data> USE_EXISTING=true make sharding-with-keywords

Command-line arguments:

• --data-dir: Specify the data directory
• --read-keywords: Provide a file containing keywords to query
• --log-trace: Enable detailed logging

GPLv3.