While secure multiparty computation (MPC) protects input data and interim outcomes, it falters when exposing input specifics through outputs. To curtail these data leaks, differential privacy proves effective. This involves injecting noise into MPC outputs before their disclosure. Thus, a discreetly collaborative method for MPC servers to generate random noise becomes indispensable.
Our MPC-DualDP protocol is designed for dual-party scenarios, where two servers collaborate to generate shared noise for the purpose of ensuring differential privacy.
Currently, we support the generation of noise following a binomial distribution, and more distributions might be supported in the future.
As for the security model, our protocol and implementation provide security and privacy against two semi-honest servers.
In terms of the protocol's efficiency, it involves only one round of communication, and if we consider a binomial distribution
MPC-DualDP guarantees that neither the adversary nor any party possesses knowledge about any portion of the noise generated, ensuring the fulfillment of the differential privacy guarantee. As a result, excessive noise injection by each party, a characteristic of other related approaches, can be eliminated. This leads to better data utility. Consequently, MPC-DualDP stands out in comparison to other similar methods. The concealed nature of noise generation for each party also guarantees the coherence of the final output in the MPC process for both parties.
- Linux
- CMake (>=3.15)
- GNU G++ (>=5.5) or Clang++ (>= 5.0)
- Python 3
MPC-DualDP depends on PPAM. You need to build the dependencies of PPAM first, and then build MPC-DualDP.
First, install NASM with apt install nasm
, or build and install NASM from source code downloaded from NASM's official page. Assume that you are working in the root directory of NASM source code.
./configure
make -j
make install
Second, build and install IPCL using the following scripts.
Assume that IPCL is cloned into the directory ${IPCL}
and will be installed to the directory ${IPCL_INSTALL_DIR}
.
cmake -B ${IPCL}/build -S ${IPCL} -DCMAKE_INSTALL_PREFIX=${IPCL_INSTALL_DIR} -DCMAKE_BUILD_TYPE=Release -DIPCL_TEST=OFF -DIPCL_BENCHMARK=OFF
cmake --build ${IPCL}/build -j
cmake --build ${IPCL}/build --target install
Third, build JSON for Modern C++ (JSON) using the following scripts.
Assume that JSON is cloned into the directory ${JSON}
.
cmake -B ${JSON}/build -S ${JSON}
cmake --build ${JSON}/build -j
At last, build MPC-DualDP using the following scripts.
Assume that MPC-DualDP is cloned into the directory ${MPC-DualDP}
.
cmake -B ${MPC-DualDP}/build -S ${MPC-DualDP} -DCMAKE_BUILD_TYPE=Release -DIPCL_DIR=${IPCL_INSTALL_DIR}/lib/cmake/ipcl-2.0.0 -Dnlohmann_json_DIR=${JSON}/build
cmake --build ${MPC-DualDP}/build -j
Output binaries can be found in ${MPC-DualDP}/build/lib/
and ${MPC-DualDP}/build/bin/
directories.
Compile Options | Values | Default | Description |
---|---|---|---|
CMAKE_BUILD_TYPE |
Release/Debug | Release | The build type. |
MPC-DUALDP_BUILD_SHARED_LIBS |
ON/OFF | OFF | Build a shared library if set to ON. |
MPC-DUALDP_BUILD_EXAMPLE |
ON/OFF | ON | Build C++ example if set to ON. |
MPC-DUALDP_BUILD_TEST |
ON/OFF | ON | Build C++ test if set to ON. |
MPC-DUALDP_BUILD_DEPS |
ON/OFF | ON | Download and build unmet dependencies if set to ON. |
MPC-DualDP further depends on Google Test. The build system will try to find these dependencies if they exist or will otherwise automatically download and build them.
Here we give a simple example to run our protocol.
To run Party A
cd ${MPC-DualDP}/build/bin
./mpc_dualdp_example 127.0.0.1 8899 127.0.0.1 8890 0
To run Party B
cd ${MPC-DualDP}/build/bin
./mpc_dualdp_example 127.0.0.1 8890 127.0.0.1 8899 1
MPC-DualDP is Apache-2.0 License licensed, as found in the LICENSE file.
This software is not an officially supported product of TikTok. It is provided as-is, without any guarantees or warranties, whether express or implied.