A Power Model for Monitoring Environmental Parameters on the Edge

Abstract

In the Internet of Things, connected devices capture real-time data which is sent to the Fog or Cloud layer for processing. The transfer of large data volumes is subject to latency and variable transfer rates. Edge computing is an emerging trend that aims at processing data nearer to its source thereby reducing data transfer and the need for continuous connectivity. The main contribution of this paper is to investigate the power requirements of a battery-powered Raspberry Pi which acts as an Edge node. The latter monitors essential environmental parameters such as air temperature, air humidity, soil moisture and light intensity. The sensors are connected to an Arduino interfaced with the Raspberry Pi. A software-based power model relative to CPU utilization is proposed to measure the power consumption of the Raspberry Pi. A Java program is used to capture and save the sensor values in a database for further analysis. Experiments demonstrate that the power usage increases linearly with CPU utilization. The proposed power model has a root mean squared error of 0.023 and 0.036 respectively when one and two Arduinos are connected. The power requirement of a single Raspberry Pi with four Arduinos is estimated to be 2.58 Watt when the Java process is running. It is also projected that a 6,500 mAh battery can power this type of environmental setup that monitors plant growth for 12 hours.