Improving Quality Using Architecture Fault Analysis with Confidence Arguments
March 2015 • Technical Report
Peter H. Feiler, Charles B. Weinstock, John B. Goodenough, Julien Delange, Ari Z. Klein, Neil Ernst
The case study shows that by combining an analytical approach with confidence maps, we can present a structured argument that system requirements have been met and problems in the design have been addressed adequately.
Publisher:
Software Engineering Institute
CMU/SEI Report Number
CMU/SEI-2015-TR-006
DOI (Digital Object Identifier):
10.1184/R1/6574250.v1Subjects
Abstract
This case study shows how an analytical architecture fault-modeling approach can be combined with confidence arguments to diagnose a time-sensitive design error in a control system and to provide evidence that proposed changes to the system address the problem. The analytical approach, based on the SAE Architecture Analysis and Design Language for its well-defined timing and fault behavior semantics, demonstrates that such hard-to-test errors can be discovered and corrected early in the lifecycle, thereby reducing rework cost. The case study shows that by combining the analytical approach with confidence maps, we can present a structured argument that system requirements have been met and problems in the design have been addressed adequately—increasing our confidence in the system quality. The case study analyzes an aircraft engine control system that manages fuel flow with a stepper motor. The original design was developed and verified in a commercial model-based development environment without discovering the potential for missed step commanding. During system tests, actual fuel flow did not correspond to the desired fuel flow under certain circumstances. The problem was traced to missed execution of commanded steps due to variation in execution time.