Flexible Devices for Arctic Ecosystems Observations

  • Lukasz Sergiusz Michalik Universitetet i Tromsø - Norges arktiske universitet
  • Otto Anshus Universitetet i Tromsø - Norges arktiske universitet
  • John Marcus Bjørndalen Universitetet i Tromsø - Norges arktiske universitet


Devices for observing the environment range from basic sensor systems, like step-counters, through wild-life cameras, with limited processing capabilities, to more capable devices with significant processing, memory and storage resources. Individual usage domains can benefit from a range of functionalities in these devices including flexibility in prototyping, on- device analytics, network roaming, reporting of data, and keeping the devices and services available in spite of failures and disconnections. The problem is that either the devices are too resource limited to support the range of functionalities, or they use too much energy.

An important usage domain is COAT – Climate-Ecological Obser- vatory for Arctic Tundra. Presently, best practice includes deploying wild-life cameras in the Arctic Tundra, and visiting them to manually collect the recorded observations. This is a problem because such devices can only be rarely visited, and manual approaches to fetching data and storing it do not scale with regards to number of cameras, handling of human mistakes, and with freshness of observations.

We present a prototype for observing the environment composed of a general purpose computer, a Raspberry PI, in combination with an ARM-based microcontroller. The combination enables us to create a more energy efficient prototype while supporting the needed functionality.

The prototype improves on currently applied methods of observing the Arctic tundra. The prototype automatically observes the arctic tundra through camera, humidity and temperature sensors. It monitors itself for failures. The data is stored locally on the prototype until it can be automatically reports to a backend service over a wireless network.

We have conducted experiments that show that task scheduling can reduce power consumption, and we identify some additional points that need to be addressed before we can run the device for long periods on battery power.