All projects
03 / 04

GeoMCU Sensor Platform

A dashboard for operating networked GeoMCU sensors and analyzing live or recorded vibration data. Live MQTT streams, wired devices, and CSV/Excel replays use shared parsing, buffering, feature extraction, and health-monitoring interfaces.

Real-time systems · Python · NetworkingUndergraduate research · UbiHealth Lab
GeoMCU Sensor Dashboard showing network status, live measurements, and signal charts
01Ingestion

Connection adapters normalize live and recorded sensor inputs.

MQTT, wired, Bluetooth, and file-based readers convert incoming data into a shared sensor-packet model. This keeps device-specific connection logic outside the buffering and analysis layers.

02Processing

One processing pipeline handles live streams and replay files.

Parsed packets enter a shared buffer used by signal-feature calculations, stream-quality checks, and visualization. Replay data follows the same path as live sensor data, which keeps algorithm behavior consistent across collection and analysis workflows.

03Monitoring

The dashboard reports sensor state and stream quality with the signal data.

Per-sensor views expose packet counts, buffer size, last-packet delay, amplitude, peak-to-peak value, RMS, standard deviation, dominant frequency, and connection state. Session data and run summaries are stored for later analysis.

Architecture

Data and control flow

01GeoMCU array
02Connection router
03Shared buffer
04Processing pipeline
05Research dashboard
Tools and technologies
PythonMQTTStreamlitNetworkingSignal processing
Next project

Rap Market Index