Automated Detection of Architecture Patterns in MBSE Models

The evaluation of a system’s architecture is an essential process within the systems engineering lifecycle. Commercially available model-based systems engineering (MBSE) tools, when combined with standards-based architecture modeling languages, provide a means through which architecture information can be expressed graphically and formally in a machine-readable format; this format can be leveraged in order to improve the system architecture evaluation process. The authors propose an automated, repeatable method for detecting patterns of interest embedded within an MBSE model. This novel method uses a heuristically guided set of similarity measures that depend on textual and graphical content of a model. The proposed method has been implemented for architectures developed in IBM’s Rational Rhapsody, and No Magic Inc.’s MagicDraw, and has proven to be able to identify six well-established patterns: Adapter, Bridge, Composite, Façade, Observer, and Proxy. This automation has the potential to produce cost and time savings for the evaluation process and to add an additional degree of rigor and completeness to an architecture evaluation.