Skip to content

Latest commit

 

History

History
228 lines (207 loc) · 10 KB

README.md

File metadata and controls

228 lines (207 loc) · 10 KB

MPC to 2PC Converter

The project securely converts MPC additive shares to 2PC additive shares.
Conversion is implemented by sending masked values of all MPC shares to one of the two parties, and sending the masks to the other party.
By summing their inputs, the two parties get 2PC shares of the original MPC values.

Directory Structure

  • proj - main project directory with source files
  • proj/scripts - python scripts for end-2-end testing of the project along with mock testing of actors.
  • unitest - MPC data files used for testing. We have a folder structure that contains 0p testing, 3p testing and 5p testing files. We have a separate dummy material file P0-sample_matrix_triple.txt inside the 0p_test folder that is used in database testing files.

Build instructions

Dockerized environment

  • Clone the Converter:

    git clone git@github.com:Crypto-TII/FANNG-MPC-Converter.git

  • In the same directory where you cloned the repo copy the docker files from inside the repo.

    cp ./FANNG-MPC-Converter/docker-files/development-environment.dockerfile ./
cp ./FANNG-MPC-Converter/docker-files/docker-compose.yaml ./

  • In the same directory as your repo, make the database folders:

    mkdir postgresql mysql

  • In the end, your working directory would have 5 things: FANNG-MPC-Converter, mysql, postgresql, development-environment.dockerfile and docker-compose.yaml

  • Now in the docker-compose.yaml file you need to add the path from root for the mysql, postgres and FANNG-MPC-Converter repo. It is instructed inside the docker-compose on where to add the path.

  • Now build and run the docker:

    docker-compose build --no-cache   # This is only needed once
docker-compose up

  • Now go inside the docker container:

    docker ps

  • Copy the container id of the development-environment container

    docker exec -it CONTAINER_ID /bin/bash
cd /home/MPCTo2PCConverter/proj
./build.sh

Build instruction with TaasLib

  • Clone the Converter as above!

  • Clone the Taaslib repo (This will be available in the future!)

  • Copy the combined docker-compose-taas-converter.yaml file

    cp ./FANNG-MPC-Converter/docker-files/docker-compose-taas-converter.yaml ./
cp ./FANNG-MPC-Converter/docker-files/development-environment.dockerfile ./

    Add path as instructed in the above file!

  • Run the docker-compose file

    docker-compose -f docker-compose-taas-converter.yaml up

  • Set taas = True in the two configuration files here: file1 and file2

  • Now inside the MPCTo2PCConverter folder run the following:

    python3 db_scripts/create_db_mysql.py

  • Also in this file, please remove the material type

  • Execution of the tests will remain the same as listed below and in this readme

Functional Testing

  • First step, copy one of the config files from the unittest section and place it in the config folder, as shown below:
cp unittest/3p_test/config.yaml proj/config/
  • Four unittest executables are created in the build directory: tests, tests_mysql, tests_postgresql and grpc_test. 
    cd /home/MPCTo2PCConverter/proj/build
./tests
./tests_mysql
./tests_postgresql
./grpc_test
  • Running these executables returns an error code of 0 for success, and an error code of 1 for failure.
  • One of the tests (the HTTPS communication) depends on SSL keys which are created during the build. If the files (3 files of the form: test*.pem) are missing, rerun the build command
  • Note, that a MySQL server needs to run for the tests_mysql unitests.
  • Note, that a PostgreSQL server needs to run for the tests_postgresql unittests.
  • To run the end to end testing of the new Converter service, follow this README. [Note: MySQL and a PostgreSQL server needs to be running for running the above scripts.]

Configurations for Testing Converter functionality

For testing we provide three modes:

  1. 0p testing: Trusted dealer setup with one dealer. The configuration parameters for this mode can be adjusted in this config.yaml file.

  2. 3p testing: 3 dealer setup and the configuration parameters for this mode can be adjusted in is this config.yaml file.

  3. 5p testing: 5 dealer setup and the configuration parameters for this mode can be adjustted is this config.yaml file.

For detailed instructions, go to this README.

For added reference, the configuration for model owner's database files can be changed as below inside the 5p_test/config.yaml file if your testing mode is 5p.

mo_mysql_host: "localhost"
mo_mysql_port: 3306
mo_mysql_user: "root"
mo_mysql_password: ""
mo_mysql_database: "model_owner"
mo_postgres_host: "localhost"
mo_postgres_port: 5432
mo_postgres_user: "postgres"
mo_postgres_password: ""
mo_postgres_database: "model_owner"

You have the added option to change database configurations for all actors: model_owner, super_dealer, client, dealer_0 to dealer_9 along with their rest or grpc server configurations and other parameters in the same file.

Individually starting the actors

To individually start our actors, we need to follow a specific order so that the functionalities can work in the way they are intended.

  1. Firstly, we create the keys and certificates that will be used by our actors to launch their servers!
cd /home/MPCTo2PCConverter/proj/
python3 scripts/create_keys.py
  1. Now we build the project:
mkdir build
./build.sh
  1. Next, the user has to move to the build directory inside the proj folder:
cd build
  1. Start Super Dealer
./super_dealer --id 0
  1. Start Dealer.
./dealer --id 1
  1. Start Model Owner
./model_owner --id 0
  1. Finally, to start the Client!
./client --id 0

Client as an API

Client has two functions that can be separately invoked as long as the three actors model_owner, super_dealer and dealer are active.

This endpoint initializes the client with the provided parameters.

    Client client(Parameters); # Initialize Client!
    Description: Initializes the client with the provided configuration parameters.
    Parameters:
        client_db_host (string): The host of the client's database.
        client_db_user (string): The username for accessing the client's database.
        client_db_password (string): The password for accessing the client's database.
        client_database (string): The name of the client's database.
        client_db_port (int): The port of the client's database.
        https (bool): Indicates whether HTTPS is used.
        db_system (int): The type of database system MySQL: 0 & PostgreSQL: 1.
        mo_host (string): The host of the model owner.
        mo_port (int): The port of the model owner.
        mo_cert (string): The certificate of the model owner.
        dealer_rest_hosts (vector of string): The hosts of dealer REST services.
        dealer_rest_certs (vector of string): The certificates of dealer REST services.
        dealer_rest_ports (vector of int): The ports of dealer REST services.

Function 1: get_batch_id

This endpoint retrieves the batch ID for a given model ID.

    Description: Retrieves the batch ID for a given model ID.
    Parameters:
        model_id (int): The ID of the model for which the batch ID is requested.
    Response:
        batch_id (string): The batch ID associated with the provided model ID.
        total_shares_per_batch (int): Total number of shares for the batch.

Function 2: get_shares_seed_and_mac_mask

This endpoint retrieves the shares seed and MAC mask for a given batch ID and dealer indices.

    Description: Retrieves the shares seed and MAC mask for a given batch ID and dealer indices.
    Parameters:
        batch_id (string): The batch ID for which the shares seed and MAC key mask are requested.
        dealer_indices (vector of int): The indices of dealers from which to retrieve the shares seed and MAC mask.
    Response:
        shares_seed (vector of string): The shares seed associated with the provided batch ID and dealer indices.
        mac_key_mask (vector of string): The MAC key mask associated with the provided batch ID and dealer indices.

Adding more actors

To add more actors, the user needs to add configurations for the specific actors be it dealers, model_owners, clients or super_dealers in the main config.yaml file. For instance if the user wants to work with more dealers, they just need to change the num_dealers parameter by the additional number of dealers being added and inside the dealers configuration add the configuration for the new dealer with its id information.

dealers:
  - id: 10 # We have 10 dealers already starting from 0 to 9.. so this is how the user would add an extra dealer's configurations
    host: # Add information
    port: ......
    mysql_host: ......
    mysql_port: ......
    mysql_user: ......
    mysql_db: ......
    mysql_pwd: ......
    postgres_host: ......
    postgres_port: ......
    postgres_user: ......
    postgres_db: ......
    postgres_pwd: ......
    rest_host: ......
    rest_port: ......
    rest_cert: ......
    rest_privkey: ......
    grpc_cert: ......
    grpc_privkey: ......
    https: ......
    db_engine: ......
    model_id: ......
    num_dealers: ......
    max_share_size: ......
    dealer_ids_str: ......

DB Testing

For database testing in MySQL and PostgreSQL in dockerized environment, refer to this README.

Generating API Swagger docs

The proj/scripts folder contains a file server.py which implements the skeleton of the API (returning dummy results) as a FastAPI http server. See README in the script folder for details.