Protocol Architecture & Technology

System Architecture

The Pandacea ecosystem is comprised of several interoperable layers designed for security, scalability, and user control. The diagram below illustrates the high-level components and their primary interactions.

Logical Architecture Diagram from the System Design Document.

Component Breakdown

• MyData Agent (Earner): A user-controlled application that manages data sources, enforces user policies, and acts as a P2P node for data leasing and privacy-preserving computation.
• Buyer-Side Agent (Spender): A developer-focused application for discovering data products, initiating leases, and receiving results.
• P2P Network (libp2p): The communication backbone for agent discovery and negotiation, using a Kademlia DHT for decentralized node lookup.
• Storage Layer (IPFS/Filecoin): Decentralized storage for data payloads, ensuring content-addressability and persistence.
• Privacy Layer (PySyft): The framework for executing remote data science tasks, focusing on federated analysis and statistics on extracted features.
• Settlement Layer (Polygon PoS): The public blockchain for recording all value-based transactions, enforcing agreements via smart contracts, and distributing royalties.

Core Data Models (MVP Focus)

This section defines the initial data structures for the MVP, focusing on the prioritized data types for the autonomous warehouse logistics market.

DataProduct Schema

This is the public-facing listing an Earner publishes to advertise their available data.

{  
  "schemaVersion": "1.1",  
  "productId": "did:pandacea:earner:123/abc-456",  
  "ownerDid": "did:pandacea:earner:123",  
  "name": "Novel Package 3D Scans - Warehouse A",  
  "description": "High-fidelity point cloud data...",  
  "dataType": "RoboticSensorData",  
  "keywords": ["robotics", "3d-scan", "lidar"],  
  "sampleCid": "bafybeig...xyz"
}

MVP Data Type: RoboticSensorData

The schema for the actual data payload referenced by a DataProduct.

{
  "schemaVersion": "1.1",
  "captureTimestamp": "2025-07-12T19:50:00Z",
  "sensorType": "LIDAR",
  "sensorModel": "Velodyne Puck VLP-16",
  "dataFormat": "PCD_BINARY",
  "payloadCid": "bafybeig...abc",
  "metadata": {
    "warehouseId": "ATL-01",
    "taskContext": "novel_package_scan",
    "dimensions_mm": { "x": 350, "y": 210, "z": 155 }
  }
}

Economic Model & Market Stability

The Pandacea marketplace is governed by a set of transparent, DAO-controlled economic mechanisms designed to foster a healthy market and prevent a "race to the bottom".

• Pandacea Data Valuation Formula (PDVF): A heuristic-based model used at launch to calculate a fair price based on data type, rarity, source quality, and use case.
• Dynamic Minimum Pricing (DMP): A protocol-enforced floor price for all data leases, calculated by a smart contract to provide a defensive baseline of compensation for all Earners.
• Reputation-Based Royalties (RBR): The protocol's 30% royalty pool is distributed to Earners based on a weighted score of their transaction volume and on-chain reputation, creating a powerful meritocratic incentive for quality.

Security & Privacy by Design

Security is not an afterthought; it is a prerequisite. The protocol is designed with a "trust-by-proof" privacy model and robust security considerations.

• Verifiable Consent: Every data lease requires a user to cryptographically sign a Verifiable Credential (VC), creating an immutable, auditable trail of consent that forms the basis of the protocol's legal "safe harbor".
• Privacy-Enhancing Technologies (PETs): We integrate OpenMined's PySyft to enable federated learning and secure multi-party computation, ensuring that raw data is protected.
• Threat Modeling: Our development is guided by a comprehensive Security Threat Model using the STRIDE methodology to proactively mitigate risks like agent impersonation, data tampering, and economic exploits.