A visual programming approach based on domain ontologies for configuring industrial IoT installations

Abstract

Industrial products tend to be customized by third-parties for different use cases. This is currently supported by adding external Programmable Logic Controllers (PLC) to installations. As IoT software ecosystems become wide-spread, the need for such hardware will decrease. However, removing the controllers opens the challenge of configuring distributed industrial installations. This paper argues for modeling application requirements on industrial installations so that they can be evaluated automatically based on information about targeted installations. GIMLE, a visual language for modeling application requirements using expressive domain knowledge, is proposed. GIMLE enables modeling requirements on physical features of installations, which hasn't received significant attention in the related work. A study with domain experts is used to reflect on the proposed process. The scalability of the visual language is evaluated using a model for a real-world application. The support for reuse of requirements enabled us to build models that can add support for new installations without significant changes and with a slower increase in size the more components can be reused.