Localizing Multiple Faults in Simulink Models

摘要: As Simulink is a widely used language in the embedded industry, there growing need to support debugging activities for models. In this work, we propose an approach localize multiple faults Our builds on statistical and iterative. At each iteration, identify resolve one fault re-test models focus localizing that might have been masked before. We use decision trees cluster together failures satisfy similar (logical) conditions model blocks or inputs. then present two alternative selection criteria choose more likely yield best localization results among clusters produced by our trees. Engineers are expected inspect ranked list obtained from selected faults. evaluate 240 multi-fault three different industrial subjects. compare with baselines: (1) Statistical without clustering, (2) State-of-the-art clustering-based debugging. show significantly reduces number of engineers order all faults, when compared baselines. Furthermore, approach, less performance degradation than baselines increasing underlying