Metrics for verification and validation of architecture in powertrain software development

Abstract

The key challenges to software development in the powertrain domain are an exploding complexity, short development cycles and increasing quality and safety requirements. Architecture design addresses these challenges by structuring development over the entire lifecycle. Being a work product of the very early stage of development, architecture design errors lead to rework and higher development costs at any following development step. Verification and validation on architecture level leads to significant quality increase and reduces development risks. To approach this in a structured manner, the authors propose automated metric based evaluation of the software architecture. Metrics are identified and developed based on the Goal Question Metric approach. Quality goals for the architecture are derived from international standards, organizational goals and domain specific requirements. Questions are developed based on the goals to derive metrics for the evaluation of software quality criteria. The metrics developed are integrated into the software development process. At the end, application examples from series hybrid control software projects are shown. By the automated, metric-based approach so far extensive manual reviews and scenario analyses are supported by objective and measurable criteria. As an added value to state-of-the-art development strategies, frontloading of the costs is achieved by early evaluation of the software architecture.