Edge Process Management: A case study on adaptive task scheduling in mobile IoT

Abstract

Service-oriented standards and architectures provide a useful way of interfacing Internet of Things (IoT) devices with each other and outside applications, while managing the IoT-related automation and decision-making. Specifically, Business Process Management Systems (BPMS) and its’ standards (e.g. BPMN 2.0) help model the composition of devices and tasks involved in an IoT workflow. The conventional centralized BPMS architecture, where remote servers handle decision-making and orchestration, does not fit IoT paradigms such as Edge- and Fog Computing, where adaptability and de-centralized processing is required. Edge Process Management (EPM) aims to adapt BPMS to such requirements. For instance, in case the users smartphone is executing processes involving communication with nearby fog servers, the process execution engine needs to consider user mobility and available fog servers to ensure energy-efficient and adaptive edge process execution. We propose an EPM system architecture featuring a delay-aware fog server selection scheme that optimizes cost-efficiency of edge processes based on run-time context factors such as user movement, fog server load and location. We demonstrate a prototype implementation using this scheme alongside existing BPMS software and evaluate the performance of the server selection with simulated experiments, showing how interrupted connections are avoided and energy efficiency is improved.