CSP at the Cyber-Physical Edge

Abstract

Today, to do ground-based in-situ observations of the arctic tundra, researchers carry wild life cameras and other observation units into the field, manually configure the devices while on the arctic tundra, and fetch the collected data several months later. This approach does not scale. Instead, observing and reporting of data must be automated using a distributed wireless network of autonomous observation units.<p>

<p>We present the basic hardware and software architectures of the Distributed Arctic Observatory (DAO) observation units. A DAO observation unit is composed of both heavy and light computer cores. The idea is to use the heavy cores for resource demanding tasks and then power them off and shift the workload to light cores when possible in order to increase energy efficiency and extend battery life.<p>

<p>We report on initial thoughts and experiences in applying a CSP network on an observation unit to ease development of advanced functionalities, while still achieving energy efficiency for observation units.<p>

<p>In order to rapidly develop prototype systems and learn from them, we have composed observation units with a combination of Raspberry Pi computers as the heavy cores, and Arduino, Nucleo, and Sleepy Pi microcontrollers as the light cores.<p>

<p>This is work in progress.