The System

Threelayers.Oneunbreakableproof.

VerdiChain is built on three tightly integrated layers — physical hardware, machine learning validation, and blockchain minting. Each layer makes the next one trustworthy. Together, they eliminate fraud entirely.

Hardware LayerML LayerBlockchain Layer
Layer 01 — Hardware

The sensor is the source of truth.

Every VerdiChain deployment begins with a physical sensor node placed in or near the green space being credited. The node is a Raspberry Pi-based unit with a suite of environmental sensors, costing approximately $40 in components and designed for outdoor durability and low-power continuous operation.

Soil Carbon Sensor

Measures soil organic carbon concentration as a proxy for long-term biomass sequestration. Readings logged every 15 minutes.

CO₂ Flux Monitor

Tracks atmospheric CO₂ concentration differential above and at ground level to quantify net gas exchange.

Temperature & Humidity Logger

Environmental context data used by the ML model to normalize readings and cross-validate against regional climate baselines.

Biomass Growth Proxy

Correlated indices from soil moisture, conductivity, and root-zone resistance to estimate above-ground biomass growth indirectly.

All sensor data is cryptographically signed with a device-unique key at source. Unsigned data is rejected before it ever reaches the ML pipeline.

Sensor Node
Layer 02 — Machine Learning

The model decides. The model cannot be bribed.

Raw sensor data is meaningless without validation. Our ML pipeline performs three critical functions: anomaly detection to catch spoofed or faulty data, sequestration estimation to quantify CO₂ absorbed, and threshold confirmation to trigger minting only when the evidence is statistically conclusive.

01ML

Data Ingestion & Cleaning

Sensor packets arrive signed. The pipeline verifies signatures, checks for dropout patterns, and discards any anomalous readings before processing begins.

02ML

Spoofing & Anomaly Detection

A trained classifier checks reading sequences against known-good baselines for the biome type. Unnatural consistency, stepwise jumps, or sensor flatlines trigger an immediate fraud flag.

03ML

Sequestration Estimation

Validated readings feed a regression model trained on 50M+ correlated data points from peer-reviewed field studies to estimate net tonnes of CO₂ sequestered over the measurement period.

04ML

Threshold Confirmation

When the estimated sequestration crosses a configurable minimum threshold (default: 0.1 tonne CO₂e), the ML pipeline generates a signed confirmation payload and dispatches it to the smart contract.

Layer 03 — Blockchain

Minted on-chain. Owned forever.

The signed ML confirmation payload triggers an Ethereum smart contract that mints an ERC-721 NFT token representing one verified carbon credit. The token metadata includes the sensor node ID, GPS coordinates, measurement window, sequestration estimate, ML model version, and a hash of the raw data bundle — all permanently on-chain.

Token StandardERC-721 (NFT)
NetworkEthereum (Mainnet + Sepolia Testnet)
Mint TriggerML confirmation payload (signed)
MetadataNode ID, GPS, timestamp, CO₂e estimate, data hash
VerificationPublic QR code → Etherscan + VerdiChain explorer
TransferStandard ERC-721 transfer to buyer wallet
Blockchain
Verification

Anyone can verify. No account needed.

Every minted credit ships with a QR code that resolves to the VerdiChain public explorer. Scan it and see the full proof chain: sensor readings, ML validation log, minting transaction hash, GPS location, and current ownership. Regulators, buyers, and the public all see the same immutable record.