Edge-Native Earthquake Detection

A supercomputer-trained neural network.
Optimized for a $10 microcontroller.
Life-saving early warning at scale.

EdgeQuake runs a custom TensorFlow Lite phase-picking model directly on the ESP32-S3 — no cloud round-trip, no hub, no latency tax. Deploy hundreds of nodes across a city and build a real-time early-warning mesh that gets smarter as it grows and scales.




<50ms Edge Detection Latency
ESP32-S3 Compute Core
MQTT Cloud Transport
v0.2 Hardware Prototype
How It Works

From ground motion
to cloud alert in milliseconds.

EdgeQuake runs the detection algorithm locally so the most critical seconds aren't lost to network round-trips.

01 — Sense

Ground Motion Captured

An onboard MEMS accelerometer continuously samples ground acceleration at high frequency, feeding a real-time signal buffer.

📡
02 — Detect

On-Device Phase Picking

The ESP32-S3 runs a lightweight STA/LTA or ML-based phase picker, identifying P-wave and S-wave arrivals without cloud dependency.

03 — Report

Metadata to the Cloud

Only compact event metadata — arrival time, amplitude, phase classification — is published via MQTT, keeping bandwidth minimal.

☁️
04 — Warn

Network-Scale EEW

Aggregated reports from distributed nodes let backend systems triangulate epicenters and dispatch warnings ahead of destructive shaking.

🔔
The Device

Purpose-built for the seismic edge.

Unlike repurposed development boards, EdgeQuake is designed from the ground up for always-on seismic monitoring at low cost and low power.

True Edge Computing

Phase detection runs entirely on-device. Alerts fire even when internet connectivity is intermittent — critical for post-event scenarios.

📶

Wi-Fi + Cloud Native

Built around the ESP32-S3 with integrated Wi-Fi. MQTT over TLS delivers structured event payloads to any broker or cloud endpoint.

🔩

Open & Hackable

Standard USB-C power, accessible boot and reset controls, and exposed GPIO. Built for researchers, not black-box deployments.

💰

Dense Network Economics

Low per-unit cost makes hundred-node urban deployments viable — the density that meaningful early warning demands.

EdgeQuake v0.2.1 prototype board
MCU
ESP32-S3
Connectivity
Wi-Fi 802.11n
Protocol
MQTT / TLS
Power
USB-C 5V
Version
v0.2.1 Proto
Form Factor
85 × 65 mm
Landscape

How EdgeQuake stacks up.

Grillo and Raspberry Shake have proven the market. EdgeQuake advances the edge-native detection model.

Capability EdgeQuake Grillo Raspberry Shake
On-device phase picking
Cloud event metadata streaming
Sub-$100 target unit cost
Designed for dense urban arrays
Standard Wi-Fi (no hub required) Some models
Geoscience-grade sensor option Roadmap
Applications

Where EdgeQuake belongs.

Any environment that benefits from dense, real-time seismic sensing at affordable scale.

Early Warning

Urban EEW Networks

Deploy hundreds of nodes across a city to build redundant P-wave detection meshes, buying critical seconds before S-wave arrival.

Research

Seismological Studies

Dense low-cost arrays for aftershock monitoring, induced seismicity research, and site response characterization in complex basins.

Infrastructure

Critical Facility Alerting

Hospitals, data centers, and utility control rooms gain automated shake alerts to trigger protective protocols before shaking arrives.

CCUS / O&G

Induced Seismicity Monitoring

Real-time event detection for carbon capture, wastewater injection, and geothermal operations under regulatory monitoring requirements.

Education

University Deployments

Bring real seismological data into classrooms and undergraduate research programs at a price point that fits departmental budgets.

Community

Citizen Science Networks

Grassroots sensor arrays in earthquake-prone regions where official monitoring infrastructure is sparse or underfunded.

Get Involved

We're early stage.
Shape what comes next.

EdgeQuake is actively seeking beta partners, researchers, and early adopters. Tell us about your use case and we'll be in touch as hardware availability opens up.

Current Status

Hardware v0.2.1 prototype complete. Firmware under active development. Targeting beta hardware program in 2025–2026. Priority access for research institutions and municipalities.

You're on the list. We've received your information and will reach out as EdgeQuake availability opens up. Thank you for your interest.

No spam. We'll only reach out about EdgeQuake availability and major milestones. Unsubscribe anytime.