Computational Offloading for the Industrial Internet of Things: A Performance Analysis

Abstract

The challenge of minimizing mission response times and the energy consumption of computationally-constrained IoT devices, is becoming increasingly important in Industry 4.0 environments, where mission-critical tasks require real-time and latency-sensitive processing. In this context, the computational offloading paradigm becomes an attractive alternative for utilising remote, computationally capable resources placed at the network Edge, to execute demanding applications. In this paper, an analysis of the main options for task execution (ondevice, Edge and Cloud) is performed, under different sensing and actuation scenarios. These results outline the need of an intelligent decision making mechanism to dynamically alternate between the available execution environments, as a response to the varying networking conditions.