Self-Healing for Distributed Workflows in the Internet of Things

Abstract

Distributed process execution is a scalable solution for implementing workflows that deal with many devices in the Internet of Things (IoT). In hierarchical network structures the execution of activities and subprocesses can be moved closer to the computing edge onto specialized devices while still preserving data security. However mobility, resource constraints and varying connectivity of devices lead to new sources of errors that jeopardize process execution. In this work we show the application of the MAPE-K feedback loop as a framework for self-healing for distributed processes. Additional context data is incorporated into the execution to analyze the state of process resources, detect errors and find compensations. Experiments show an increased success rate of process executions in a smart home use case involving service robots. The framework proves to increase resilience of Business Process Management systems while being extensible and applicable in the field of IoT.