A Software Fault Tree Metric

Abstract

Analysis of software fault trees exposes hardware and software failure events that lead to unsafe system states, and provides insight on improving safety throughout each phase of the software lifecycle. Software product lines have emerged as an effort to achieve reuse, enhance quality, and reduce development costs of safety-critical systems. Safety-critical product lines amplify the need for improved analysis techniques and metrics for evaluating safety-critical systems since design flaws can be carried forward though product line generations. This paper presents a key node safety metric for measuring the inherent safety modeled by software fault trees. Definitions related to fault tree structure that impact the metric's composition are provided, and the mathematical basis for the metric is examined. The metric is applied to an embedded control system as well as to a collection of software fault tree product lines that include mutations expected to improve or degrade the safety of the system. The effectiveness of the metric is analyzed, and observations made during the experiments are discussed