Post-quantum Oblivious Transfers

Implementations of post-quantum oblivious transfers under emp-ot framework, including:

• Our new Naor-Pinkas-like 1-out-of-2 OT protocol based on SABER;
• A Simplest-like 1-out-of-2 OT protocol based on SABER;
• Masny-Rindal 1-out-of-2 OT protocol based on SABER;
• Masny-Rindal 1-out-of-2 OT protocol based on CRYSTALs-Kyber.

The first two protocols are our new protocols inspired by Naor-Pinkas OT and Simplest OT. In our paper, we mainly focus on the first Naor-Pinkas-like construction. The other two are following the Masny-Rindal construction from uniform key agreement, as baselines.

Dependency

Our implementation uses emp-tool. Here is how to install it, more details can be found in their README.

1. wget https://raw.githubusercontent.com/emp-toolkit/emp-readme/master/scripts/install.py
2. python install.py --deps --tool
   1. You can use --ot=[release] to install a particular branch or release
   2. By default it will build for Release. -DCMAKE_BUILD_TYPE=[Release|Debug] option is also available.
   3. No sudo? Change CMAKE_INSTALL_PREFIX.

Build

After installing the dependencies, you can build Saber-based project with the following commands

cmake -DENABLE_SABER=ON  -DENABLE_KYBER=OFF .
make
./run ./bin/test_saber

where test_saber is our test program for the SABER-based OTs, including NPOTs, Simplest OTs, and MR OTs. Additionally, it also inlcudes some CDH-based classical OTs as contrast. The output should be like:

connected
connected
128 COOTs (average over 1000 runs):	13.921 ms
128 COOTs (average over 1000 runs):	13.91 ms
128 NPOTs (average over 1000 runs):	14.667 ms
128 NPOTs (average over 1000 runs):	14.678 ms
128 NP Saber OTs (average over 1000 runs):	8.221 ms
128 NP Saber OTs (average over 1000 runs):	8.235 ms
128 Simplest Saber OTs (average over 1000 runs):	6.349 ms
128 Simplest Saber OTs (average over 1000 runs):	6.36 ms
128 MRSaber OTs (average over 1000 runs):	18.568 ms
128 MRSaber OTs (average over 1000 runs):	18.587 ms

Similarly, you can build Kyber-based project with the following commands:

cmake -DENABLE_SABER=OFF  -DENABLE_KYBER=ON .
make
./run ./bin/test_mr_kyber

where test_mr_kyber is the test program running our Kyber-based MR OTs, still with some classic CDH ones for contrast. The output should be like:

connected
connected
128 COOTs (average over 1000 runs):	12.921 ms
128 COOTs (average over 1000 runs):	12.928 ms
128 NPOTs (average over 1000 runs):	14.866 ms
128 NPOTs (average over 1000 runs):	14.898 ms
128 MRKyber OTs (average over 1000 runs):	17.684 ms
128 MRKyber OTs (average over 1000 runs):	17.681 ms

Because there are too many collsions in the implementations of Saber and Kyber, you can only enable one of them at a time.

The tests above are running on the same machine's loopback network. Credited to emp-ot for their complete implementation, one can also test these OTs on two different machines in a real network setting, more details to use their features can be found in their README.