A Python library for important oblivious algorithms

This library implements a series of classic oblivious algorithms, covering oblivious RAM (ORAM), oblivious map (OMAP), and algorithms for oblivious graph processing. It is built for client/server deployment scenarios in which the client is allowed to execute non-obliviously.

We list the maintainer’s related publications, which motivated the development of this library, and then provide a detailed overview of all included algorithms.

• Enabling Index-free Adjacency in Oblivious Graph Processing with Delayed Duplications (Full Version)
• Towards Practical Oblivious Map (VLDB 2025)

Oblivious Graph Methods

• GraphOS (VLDB 2024), implemented here.
• Grove with delayed duplications, implemented here.

ORAM Methods

• Path ORAM (CCS 2013), implemented here.
• Recursive Path ORAM (ASIACRYPT 2011), implemented here.
• Insecure Freecursive ORAM (ASPLOS 2015), implemented here. The reset_method should be set to "hard".
• Secure Freecursive ORAM with probabilistic resets (TCC 2017), implemented here. The reset_method should be set to "prob".
• DAORAM with fixed reset and de-amortized cost (VLDB 2025), implemented here.

OMAP Methods

• OMAP based on AVL (CCS 2014), implemented here. Set distinguishable_search to False.
• OMAP based on optimized AVL (VLDB 2024), implemented here. Set distinguishable_search to True.
• OMAP based on B+ tree (VLDB 2020), implemented here.
• OMAP framework over an underlying search tree (VLDB 2025), implemented here. It can be instantiated with any ORAM class in this repo and combined with any tree-based OMAP class in this repo.
• Cache-optimized AVL and B+ tree variants are implemented here and here, respectively.

Project Structure

• The demo folder contains demonstrations showing how to use sockets to set up a server and client.
• The dependency folder contains required dependencies, including socket and cryptography modules.
• The graph folder contains all oblivious graph constructions included in the library.
• The omap folder contains all OMAP constructions included in the library.
• The oram folder contains all ORAM constructions included in the library.
• The tests folder contains test cases for validating the correctness of the implementations.

How to run this code

1. Install the dependencies listed in requirements.txt.

2. Local server (for testing purposes):

   • See sample usages in tests/test_orams.py and tests/test_omaps.py.

3. Remote server:

   • Start the server on the remote machine, refer to: demo/server.py.
   • Run a client on your device, refer to: demo/oram_client.py or demo/omap_client.py.