Build a Pump Monitor for Railway Applications with Zephyr OS
summary
Device built around the nRF9160 SiP sends encoded MQTT reports over LTE-M/NB‑IoT and runs local pattern recognition on Zephyr to decide pump OK versus failure. Hardware combines temperature moisture pressure sensors and an accelerometer with power gating for the 5 V pressure sensor, 50 Hz pressure sampling, and NFC polling at 10 Hz. Deployment for Deutsche Bahn runs 36 modules 24/7, logs over 50 000 pump events, and achieves maintenance‑free battery life up to three months.
AI-written summary · about
chapters (27)
- 0:00Pump monitoring project intro
- 1:03Wastewater tank overflows problem
- 2:53Battery-powered sensor challenges
- 4:21Hybrid wastewater flow detection
- 5:03Zephyr RTOS for firmware
- 6:313 years of 24/7 results
- 7:56Edge vs cloud computing
- 9:21AI limits in flow detection
- 13:23Domain knowledge vs AI
- 14:28Real-time signal processing
- 17:45Firmware overview
- 18:06Hardware components
- 19:12Low-power design challenges
- 20:34Zephyr modules and SDK
- 22:24Power states and polling
- 23:29Energy profiling kit
- 25:12Cheap vs expensive devices
- 26:14Power gating issues
- 27:43Power gating trade-offs
- 28:47MQTT versus lightweight M2M decision
- 29:07SD card storage resilience
- 30:12Rapid prototyping lessons
- 31:18Field testing results
- 31:59Debugging and telemetry
- 32:42Water intrusion alarm
- 33:23Battery life improvement plans
- 34:29Question and answer session
related talks
31mSmart Train Stations with LoRaWAN
35mEnd-to-end IoT Development with Zephyr
36mConnecting Zephyr Devices at Scale Using Open Source Solutions
14mKeynote: Leveraging Zephyr and ML to Bring Smart Devices to Market, Faster
39mFrom a Weekend Hack, to a Future-Proof IoT Device
25mBreton Industrial Datalogger
39mFrom a Weekend Hack, to a Future-Proof IoT Device
10mLightning Talk: Open Source Fleet Management in Zephyr