Quality Assessments in MBSE: past, present and future

Zusammenfassung

Model-Based Systems Engineering (MBSE) relies on system models as primary engineering artifacts. Their quality is critical for using models as a single source of truth, for artifact generation, and for automated verification. However, increasing system complexity, interdisciplinary collaboration, and diverse stakeholder needs make it difficult to maintain model quality throughout the lifecycle of cyber-physical systems.A wide range of model quality assessment approaches has been proposed, yet the field lacks an up-to-date, structured overview of their strengths, limitations, and industrial applicability, which hinders both further research and adoption in practice.To investigate the state of the art, we conduct a Systematic Literature Review of model quality assessment approaches applicable to system model elements, covering primary studies published between 1994 and 2026. The final set comprises 80 primary studies. We complement the review with 10 semi-structured practitioner interviews to compare against current practice.The results characterize the past and present of model quality assessment in MBSE. The literature is dominated by metric- and rule-based assessments of UML models, primarily focusing on syntactic correctness and model smells. Reports on the industrial applicability of approaches, implementation effort, and required expertise are sparse in the literature. Practitioners emphasize context-dependent model element quality as the key success factor for effective assessment. We use the identified gap between state of the art and practice to derive six design requirements for the future of model quality assessments in MBSE.