A framework for zero-knowledge-based authentication as an alternative for traditional authentication mechanisms. Relying on client-server architecture, the framework ensures that passwords always remain on the client and are never passed to the server, and instead derives parameters from the credentials. The server-side stores these parameters as password-equivalent data. The authentication of registered user happens utilizing the concept of Zero-Knowledge Proofs, where the actual chosen credentials get exposed neither to the server nor the medium between the client and server. Zero-Knowledge Proofs require a computationally "hard problem" to carry out the authentication. Currently we implement a graph isomorphism-based ZKP protocol. The framework is designed such that any hard problem can be simply swapped easily.

• Python 3.10+
• Dependencies from requirements.txt
• MongoDB

Run pip install -r requirements.txt to install all dependencies once a Python 3.10+ virtual environment has been created.

Ensure that MongoDB is running before running the following commands.

Create a file with the following content

ZEROK_DB_HOST=127.0.0.1 # change as per MongoDB config
ZEROK_DB_PORT=27017 # change as per MongoDB config
ZEROK_DB_USER= # change as per MongoDB config
ZEROK_DB_PASS= # change as per MongoDB config
ZEROK_DB_NAME=zkp_users
ZEROK_BATCH_SIZE=5
ZEROK_ROUNDS=10

Now run python run_server.py to start the server. Then run python main.py to run the client.

stats.py generates statistics on the graph generation schemes and helps in determining the effectiveness of the various approaches used.

latency_test.py is used for testing the time taken for registration and authentication flows to complete and compare the traditional authentication mechanisms with ZKP-based mechanism. Testing for various latency and bandwidth scenarios was done using Mininet.

• Aayush Jain - https://github.com/Aayushjn
• Adwait Gondhalekar - https://github.com/adwaitgondhalekar