Privacy Utils Secure Aggregation

Secure aggregation protocols for privacy-preserving multi-party computation and federated learning.

Features

• Secure Multi-Party Computation: Privacy-preserving collaborative computation
• Federated Learning: Distributed model training without data sharing
• Homomorphic Encryption: Encrypted data processing
• Secret Sharing: Distributed secret reconstruction
• Masking Techniques: Differential privacy with secure aggregation
• WebSocket Communication: Real-time secure aggregation protocols
• Threshold Cryptography: Distributed key generation and signing

Installation

npm install privacy-utils-secure-agg

Usage

import {
  SecureAggregator,
  SecretSharer,
  MaskingEngine,
  WebSocketAggregator
} from 'privacy-utils-secure-agg';

// Create secure aggregator
const aggregator = new SecureAggregator({
  minParticipants: 3,
  maxParticipants: 10,
  privacyBudget: { epsilon: 1.0, delta: 1e-5 }
});

// Add participant data
aggregator.addParticipant('user1', [1, 2, 3, 4, 5]);
aggregator.addParticipant('user2', [2, 3, 4, 5, 6]);
aggregator.addParticipant('user3', [3, 4, 5, 6, 7]);

// Perform secure aggregation
const result = await aggregator.aggregate();

// Secret sharing
const sharer = new SecretSharer({ threshold: 3, totalShares: 5 });
const shares = sharer.shareSecret('sensitive-data');

// Reconstruct secret (needs threshold number of shares)
const reconstructed = sharer.reconstructSecret(shares.slice(0, 3));

// WebSocket-based aggregation
const wsAggregator = new WebSocketAggregator({
  port: 8080,
  participants: ['ws://node1:8080', 'ws://node2:8080', 'ws://node3:8080']
});

wsAggregator.start();

API Reference

For complete API documentation, see the TypeScript definitions or visit the main repository.

Architecture

The package provides both client and server components:

• Client: Participates in secure aggregation protocols
• Server: Coordinates the aggregation process
• Network Layer: WebSocket-based communication
• Crypto Layer: Homomorphic encryption and secret sharing

Security Model

• End-to-End Encryption: All communications are encrypted
• Zero-Knowledge Proofs: Verify computations without revealing data
• Differential Privacy: Add noise to protect individual contributions
• Threshold Security: No single party can reconstruct sensitive data

License

MIT

Contributing

Contributions are welcome! Please see the main repository for contribution guidelines.