Monitoring and Detecting Failures in Wide Area IoT Networks

Abstract

Low Power Wide Area Network technologies provide an infrastructure that facilitates development of wide area IoT applications. With this comes a need for a better understanding of how components in such applications should be monitored efficiently.

Through a design science research process, an IT artifact has been developed. The artifact was released into the environment at an early stage to start giving value to the application from the beginning, allow potential improvements to be discovered more easily and increase the probability of discovering unforeseen failures. The thesis shows how monitoring can be done without adding extra instrumentation at the monitored components, but by using the data (or lack of such) produced by the components to detect failures. The system also enables easy comparison of the collected data with weather forecast data to investigate possible affects weather might have on the application.

The developed system has successfully detected, identified and notified supervisors about failures in the monitored network and in itself for 2 and a half months, and has proved itself to be dependable even during unexpected failures.