BananaMesh

BananaMesh is a DePIN network converting verified human presence into token rewards. Using World ID to prove human data source and an ESP32 BLE mesh grid, it logs raw RSSI telemetry to HCS, anchoring a privacy-first data layer used to train AI models for volumetric spatial tomographic applications.

We built the hardware foundation inside a custom chassis designed in Autodesk Fusion. At its core, the node runs on an ESP32-S3 (N8R16) paired with an external 2.4 GHz antenna to capture high-fidelity RSSI fluctuations. To ensure remote portability, it's powered by a LiPo battery managed by a TP4056 charging module, stepped up via an MT3608 boost converter to maintain a stable 5V rail for the components.

The software and data architecture relies heavily on three core partner integrations to ensure absolute data integrity and decentralization:

World ID (Selfie Check): Serves as our cryptographic anti-bot gate. Users verify human liveness before their device signal is whitelisted by the mesh network. This transforms standard BLE telemetry into a Sybil-resistant, provably human dataset, eliminating synthetic data fraud.

Hedera Consensus Service (HCS): Functions as our immutable storage data layer. The multi-node RSSI telemetry vectors are encrypted locally and streamed directly to HCS. This guarantees sub-second finality and absolute proof against data alteration or retrofitted manipulation.

X402 Protocol: Powers our privacy-first monetization engine. By using a partial-reveal algorithm natively supported by X402, we can granularly segment the volumetric spatial datasets. This allows the network to sell specific subsets of the RF soundscape to AI training buyers without exposing the broader environment or scattering user identities.