A fully functional IoT application for temperature and humidity sensing and data transfer over SSL-encrypted MQTT. Client certificate is not required.
Deploy it on several devices to collect (and store) data from multiple locations.
The app targets all ESP8266 and ESP32 boards. It has been tested with:
- Wemos WiFi-ESP8266 DevKit
- D1 Mini Pro Based on ESP8266EX
- ESP32 DevKit
Use your private or any public MQTT broker.
Includes an (optional) "accumulator" client, which collects data into a MySQL database and bins older data from previously disconnected instances.
The app and the client were built and tested on Linux. I expect the app to compile on Windows and MacOS (because the API is architecture-agnostic) and the client, if executed as a standalone script, to work as well, provided that its prerequisites are met (because it is python script). Only the chosen deployment method using systemd is linux-specific.
- One or more ESP8266- or ESP32-based boards,
- One sht3x sensor and 4 jumper cables per board;
- Wifi connection (and access to Internet if using a LAN-external broker)
- 5V or 3.3V power source (5V, if applicable for a board, works better with WiFi).
This set-up is sufficient to build the app but additional items might be required depending on deployment decisions:
- a rig for the MQTT broker (Raspberry Pi, Cloud, ...)
- device(s) to read/display data (a smartphone MQTT app, the included client, ...)
This app has to be built using a board-specific Software Development Kit:
- ESP IDF for ESP32-based boards,
- ESP8266 RTOS SDK for ESP8266-based boards.
For building the (optional) client the following are required:
- Python (see
client/requirements.txtfor the list of necessary modules) - MySQL or similar. The client must have full access to one database.
- Systemd (optional) is used as a default deployment method but it is possible to run the script as a standalone.
To deploy to a dedicated MQTT broker, one has to be able to setup and run a server.
A public MQTT broker (such as this one) requires neither system administration competences nor setting up, and could be a viable alternative, at least for testing purposes.
To enjoy the outputs of this app, a MQTT client is required. The easiest is to use a MQTT client on a smartphone (just search on the market, I'm sure there are several options available).
This app has to be built using a board-specific SDK. The roadmap for the SDK installation is described here, for ESP32, and here, for ESP8266.
Follow the relevant instructions to
- Install prerequisites for Windows, Linux, or macOS;
- Get the relevant SDK's source code (ESP IDF or ESP8266 RTOS SDK);
- Set up the toolchain;
- Set up the environment variables.
Clone (or fork) this repository:
git clone https://github.com/rytis-paskauskas/MiniQttSensorThis projects can be built using the standard workflow for a given SDK; see here and here for ESP32 and ESP8266 boards, respectively. Although the underlying build tools are different for the two SDKs, the workflows are almost identical and differ only in semantics (using make vs. idf.py, for example).
Things could potentially get messy in the case of building for both types of boards. This project adopts several simple idiosynracies, described below, with the purpose of mitigating potential issues.
The first issue is that the sdkconfig files are incompatible between the two SDKs. The two possible approaches are
- start without a
sdkconfigfile and let the build tool generate a default one, or - use one of the default templates
sdkconfig.esp8266defaultorsdkconfig.esp32default, for examplecp sdkconfig.esp8266default sdkconfig; make menuconfig.
For the same reason, the project is configured to use a non-standard build directory, build_esp8266 for ESP8266, and build (not build_esp32, because I can't get it to work cleanly with CMake) for ESP32.
See also Makefile and the CMakeLists.txt for respective control statements.
The payload is formatted as a JSON string
{measurement:{temperature: xx.xx,humidity: yy.yy}}
(and is exactly 57 bytes long).
- Configure the "Client" section in Kconfig.
- Install python prerequisites and install the client
sudo pip install -r client/requirements.txt
# or
python -m pip install -r client/requirements.txt
# and then
install_client.sh- Start (and enable) service
systemctl --user start myqttsense_client.service
# optionally also
systemctl --user enable myqttsense_client.service- Debug/troubleshoot
myqttsense_client
systemctl --user status myqttsense_client.service
journalctl --user -f- Check out the database(s) using command line (
mysql -u USER -p PWD -h HOST) orphpmyadminor equivalent tools.
The typical output from journalctl -f --user
Mar 14 18:18:37 myhost systemd[1124]: Started MyQTTSense Linux client.
Mar 14 18:18:37 myhost /my/home/dir/.local/bin/myqttsense_client.py[38015]: [INFO] Connected to MQTT broker
Mar 14 18:18:37 myhost /my/home/dir/.local/bin/myqttsense_client.py[38015]: [INFO] Connected to SQL database
Mar 14 18:18:37 myhost /my/home/dir/.local/bin/myqttsense_client.py[38015]: [INFO] subscribing
Mar 14 18:18:58 myhost /my/home/dir/.local/bin/myqttsense_client.py[38015]: [INFO] new topic: sense/sht3x/c4:5b:be:62:87:e3 => `topic_sense_sht3x_c45bbe6287e3_20220314181857`
Eclipse Mosquitto has a command-line client that is quite convenient.
Assuming that a broker has been setup on localhost, and its public certificate copied to ca_certificates/broker.crt, it is possible to subscribe to all MyQttSense messages in verbose mode (useful for debugging) like so:
mosquitto_sub -h localhost -p 8883 --cafile ca_certificates/broker.crt -t 'sense/sht3x/#' -F %J --pretty -v- Why this sensor? What is the accuracy of measurement? See this thread for inspiration and/or possible alternatives. Don't expect too much in terms of accuracy (my rule of thumb: ±1°C, ±10% RH).
See LICENCE.
@gschorcht and @UncleRus are acknowledged for providing the sht3x driver.