📡 Receiving Weather Forecasts from 433 MHz Sensors on Raspberry Pi with RTL-SDR

From Raw Radio Signals to Elegant Grafana Dashboards

At some point, checking the weather in an app stopped being enough. I wanted to know the temperature on the balcony—not from some cloud-based API, but from my own sensors, working in real time. That’s how the idea for a fully private weather monitoring system was born.

📡 1. Capturing Signals from the Air

The foundation of this project is the RTL-SDR dongle—a small radio receiver capable of intercepting transmissions from cheap 433 MHz sensors. With the help of rtl_433, I transformed raw radio signals into readable JSON data. This is the moment when “radio magic” becomes actionable data.

🔧 2. Smart Filter: Only My Sensors, Only What Matters

I didn’t want noise in the system. That’s why I created bridge.js—a small, clever Node.js script that:

• Recognizes only my sensors (by model and ID),

• Maps them to friendly names: kitchen, balcony, garden,

• Converts temperature from °F to °C,

• Publishes clean, standardized data to MQTT (weather/<name>),

• Records the timestamp of the last signal,

• Sends notifications when a sensor’s battery starts dying.

This isn’t just a filter—it’s my private “weather intelligence layer.”

🧠 3. A Backend That Understands Data

Next, the data flows into a NestJS-based backend. I built a custom MQTT client that:

• Subscribes to weather/#,

• Receives ready-to-use JSON data from bridge.js,

• Updates a cache of recent readings,

• Aggregates data and stores it in PostgreSQL.

The backend doesn’t need to know about radio protocols or sensor models. It receives clean, standardized data—and does what it does best.

🗄️ 4. Weather History in PostgreSQL

Each sensor generates multiple readings per minute. Instead of storing everything, the backend:

• Buffers the data,

• Calculates averages,

• Saves elegant records to the weather_record table.

This keeps the database lightweight, fast, and ready for analysis.

📊 5. Real-Time Weather in My Home

Finally, everything lands in Grafana, where I created a set of dashboards:

• Temperature charts segmented by sensor,

• Humidity over time,

• A diagnostic panel (battery status, last signal),

• A quick view of recent values.

This isn’t just a chart—it’s my private “weather center,” running 24/7.

🎯 Final Result

I built a system that:

• Receives radio signals from 433 MHz sensors,

• Filters and normalizes the data,

• Publishes it to MQTT,

• Aggregates and stores it in the backend,

• Displays it beautifully in Grafana.

Zero cloud. Zero vendor lock-in. Everything runs locally, quickly, and exactly how I want.