Towards a digital Z framework based on a family of architectures and a virtual knowledge graph.

The purpose of systems engineering is to, often collaboratively and following complex workflows, analyse, design, optimize, operate, and evolve complex, cyber-physical systems. This paper proposes a vision of a general framework for the design, deployment and operation of Digital Z (where Z can be model, shadow, twin, passport, avatar...). Different Digital Zs are used, often in combination, for various purposes during systems engineering. That is why we propose a family of architectures for different Digital Zs. Each Digital Z architecture is constructed based on the engineers' goals. These goals can always be reduced to the observation, satisfaction, or optimization of some Properties of Interest (PoIs). Example PoIs are safety, and average energy consumption. We propose to have one Digital Z architecture per PoI. The different Digital Zs may be combined into an ecosystem. More variability is introduced when we zoom into the deployment of Digital Zs. Common choices for network communication such as DDS and MQTT each have their own strengths and weaknesses which must be taken into account when trying to satisfy non-functional properties such as meeting real-time deadlines. We also introduce the Modelverse, a Virtual (Federated) Knowledge Graph (VKG). It is used as a source of knowledge to aid in the construction of "experiments" which answer user's questions about PoIs. These, possibly concurrent, experiments are in essence particular Digital Z ecosystems/architectures. When the experiments provide answers, these are added to the VKG knowledge base in the form (question, experiment architecture, answer). The glue between the above is a template workflow. We sketch the above concepts by means of concrete examples and compare them with existing Digital Z definitions and frameworks such as the "5D model".