Addressing Components' Evolvement and Execution Behavior to Measure Component-Based Software Reliability

TL;DR

This paper proposes a convolution-based method to measure the reliability of component-based software by modeling individual component behavior and evolution, supporting incremental development and cost-effective reliability assessment.

Contribution

It introduces a novel convolution modeling approach that integrates component reliability and evolution, enabling efficient and incremental software reliability measurement.

Findings

The approach effectively models component reliability over time.

It supports incremental updates with minimal re-evaluation.

Experiments demonstrate its usefulness in real scenarios.

Abstract

Software reliability is an important quality attrib-ute, often evaluated as either a function of time or of system structures. The goal of this study is to have this metric cover both for component-based software, be-cause its reliability strongly depends on the quality of constituent components and their interactions. To achieve this, we apply a convolution modeling ap-proach, based on components' execution behavior, to integrate their individual reliability evolvement and simultaneously address failure fixes in the time do-main. Modeling at the component level can be more economical to accommodate software evolution, be-cause the reliability metric can be evaluated by reus-ing the quality measures of unaffected components and adapting only to the affected ones to save cost. The adaptation capability also supports the incremental software development processes that constantly add in new components over time. Experiments were con-ducted to discuss the usefulness of this approach.