Beehive sensor data

Weight, temperature and humidity measurement in a beehive using Arduino and LoRaWan (TTN).\ Deutscher Text.

To support bee keepers in monitoring their bees with minimal interference, a set of sensors will measure and transmit the current state in short intervalls. Design goals:

• Energy efficiency. Devices should be able to run on batteries and solar cells for unlimited time.
• Low radiation. I do not know whether or not radio transmissions disturbe bees. Using a low-power technology like LoRa rather than GSM minimizes the effect.
• Low cost. Small microcontrollers are very affordable and sufficient for the task.

System Overview

• Arduino (alternatively Uno + Dragino LoRa Shield, Dragino LoRa Mini Dev, or CubeCell LoRa Dev)\ A small microcontroller has enough power to measure every couple of minutes. A rechargeable battery serves as buffer to measure during night and cloudy weather.\ The device works without any human intervention except a button to put it to a 'manual mode' when working with the beehive.

  • Several (e.g. 5) DS18B20 temperature sensors using 1-wire
  • DHT11/22 temperature and humidity sensor
  • Weight cell with HX711 converter
  • LoRaWan connection (TTN)
  • Solar power and rechargeable battery (LiPo)
  • Push-Button with LED to disable temporary (manual mode)
  • Evtl. slave devices with RS-485?

• LoRaWAN TTN-Gateway/Network/Application\ Beehives usually have no wired internet connection and high-speed connections using GSM and WLAN is avoided to keep radiation low.\ TheThingsNetwork (TTN) is a global open LoRa network. If no gateway is available within some km near the beehives, an additional gateway can easily be installed.

  • Transmits messages from the LoRa gateways to the internet, handles authorization (OTAA/ABP) and encoding/decoding of messages
  • Allows direct ThingSpeak integration (only 1 device per application with 1 channel)
  • HTTP integration allows to address any possible backend (many devices per application, see below)
  • see TTN application

• Dashboard and visualization with ThingSpeak channel\ This is the quick and simple way to display the sensor data.

  • Simple to setup and share ThingSpeak
  • Mobile app available
  • Matlab analysis possible
  • Limited to 8 fields/channel
  • No linking of multiple channels (beehives)
  • Limited user interaction (drilldown)

• Mapping and routing sensor data eg. using AWS (work in progress) \ Alternative or extension to the ThingSpeak display. Such a back-end application can be used to collect the sensor data and to feed a web application.

  • see AWS serverless
  • Forwarding to ThingSpeak channels as above, but allowing many devices/channels per TTN application
  • Additionally storing sensor docs in DynamoDb, see AWS recording
  • No restrictions of the number of fields or the way sensor data is displayed (own WebApp)

Device Hardware

Three boards have been tested:

• Arduino Uno + Dragino LoRa Shield
  • Needs 5V supply
  • Many digital pins already used by shield
  • Classic Uno size and MCU (16MHz ATMega328P, 32KB FLASH, 2KB SRAM)
• Dragino LoRa Mini Dev
  • 3.3V (possible LiPo, not tested)
  • Small size, LoRa transmitter integrated
  • Classic Arduino MCU (16MHz ATMega328P, 32KB FLASH, 2KB SRAM)
• Heltec CubeCell LoRa Dev. My favorite!
  • 3.3V (possible LiPo, integrated solar charging logic)
  • Power efficient (1W solar cell with 230mAh LiPo loads on a single sunny day and has almost 2 weeks of buffer)
  • Small size, LoRa transmitter integrated
  • Stronger MCU (48 MHz ARM M0+, 128KB FLASH, 16KB SRAM)

Used pins:

Transmitted LoRa message (binary encoded)

• The device measures about every 5 min
• Measures will be transmitted on significant changes or every 30 min (messages may get lost)
• Currently no uplink messages
• Fixed size and order of measured values
• Values are transmitted as short integer values with 2 digits (-327.67 .. 327.67)
• Reserved value to represent null (-327.68)

  "sensor": {
  "version": 0,  // command id or version
  "battery": 3.92,
  "weight": 0.37,
  "humidity": {
   "roof": 47.5
  },
  "temperature": {
   "drop": 20.68,
   "lower": 19.5,
   "middle": 19.93,
   "outer": 20.62,
   "roof": 22.2,
   "upper": 19.37
  }
  }

Checkout this project

Local installation of git and Arduino IDE assumed.\ Install the CubeCell board: https://heltec-automation-docs.readthedocs.io/en/latest/cubecell/quick_start.html (current version 1.3.0).

cd <project-root>
git clone https://github.com/joergkeller/beehive-sensor.git
cd beehive-sensor
git submodule init
git submodule update

To fetch the latest updates:

cd beehive-sensor
git stash
git pull
git submodule update
git stash pop

The project already includes all required libraries (that is the reason for the submodule commands and for the local sketchbook location).

Then

1. Create free TTN account/application/device and enter OTAA/ABP authorization codes in credentials.h
2. Create free ThingSpeak account/channel and a TTN ThingSpeak integration with the appropriate values
3. Compile/upload sketch
4. Activation with OTAA takes some time (even half an hour or so)
5. Press button to enter 'manual mode'
   • no LoRa messages sending, blinking LED instead
   • continous weight measuring for calibration, calculate offset/scale (eg. see Excel sheet)
   • for temperature compensation, measure weights at different temperatures also
   • scanning 1-wire temperature sensors one by one, note ids
   • press button again to start LoRa activation
6. Enter 1-wire sensor ids and weight calibration to calibration.h
7. Compile/upload sketch again
   • Wait for OTAA activation
   • First measure should appear in the local monitor, as TTN data and in ThingSpeak channel