Data gathered by my main environment sensor.
These data are gathered by my environment sensor using a BME280 sensor chip. The sensor board is placed in my living room, facing the front door.
Folders are named in the format of YYYY/mm, there are three JSON files in each of them, containing the temperature, air pressure, and (relative) humidity data of that month.
The pipeline used to work like this (before 2018/09):
- The BME280 sensor is mounted on a NodeMCU board based on ESP8266.
- The board reads data from the sensor chip every 30 seconds, and sends to an MQTT server (over WiFi).
- A Python script is connected to the MQTT server as a client, it transfers the data into a statsd instance.
- A Graphite database stores all data passing into that statsd instance. It is also used for minimal visualization.
- Data is exported as JSON monthly, and converted to CSV files.
The current pipeline works like this (since 2018/09):
- The BME280 sensor is mounted on a NodeMCU board based on ESP8266.
- The board reads data from the sensor and send them to LeanCloud (over WiFi).
- The board goes into deepsleep mode to save energy. A rewake signal would be set so the board can rewake itself to capture data every 10 minutes.
- Data is exported as CSV files monthly.
Well, why not?
- Temperature: Celsius;
- Pressure: hPa, in which 1hPa equals 100Pa;
- My sensor board runs 24/7, but date points will be missing from time to time in case of:
- Power outage
- Network issue
- Server maintenance
- These data files are missing too many data points:
2017/02: the sensor board firmware and relay script was buggy when I first set them up.2018/09: I switched from selfhosted MQTT-statsd-prometheus-grafana instance to LeanCloud storage.
- No need to say, timestamps are in UTC, but I live in GMT+8 region.
- Resolution:
- Before 2018/09, data was captured every 30 seconds, and resampled by Grafana when exporting to CSV files.
- Since 2018/09, data was captured every 10 minutes, and exported as is.
See LICENSE file for details.