Representation and selection of quality attribute techniques for software development process

Abstract

Background: The quality of software is determined during its development. Software products target specific software quality attributes, such as safety or performance. Development teams use various techniques to investigate, evaluate and control potential product quality problems. We call these “Quality Attribute Techniques” (QATs). An example QAT for safety is the hazard analysis technique of Fault Tree Analysis, and for performance is Stress Testing. Although QATs are widely used in practice, there is no systematic approach to represent and integrate QATs for arbitrary quality attributes within existing approaches to software process modelling and tailoring. Aims: This research aims to provide systematic approaches for better selection and integration of QATs into tailored software process models that target specific product qualities. Method: A framework to capture and present significant QAT information is derived by 1) reviewing literature to identify important characteristics of QATs for selection and process integration; and 2) using risk management as a general theory to understand how QATs function to manage product quality problems during development. A systematic selection method is then developed to support the choice of appropriate QATs for any quality attribute, across the lifecycle. The selection method is based on three perspectives: 1) risk management; 2) process integration; and 3) cost/benefit using Analytic Hierarchy Process (AHP). A literature-based evaluation is used to validate and improve the initial framework. An industry case study validates the feasibility and effectiveness of applying the framework and selection method. Results: The literature-based evaluation shows that the framework can accommodate a variety of quality attributes and that the categorisation scheme provides a way to choose QATs based on their impact in managing quality risks. The case study demonstrates that the framework and selection method provide a more methodological and effective approach to choose QATs for projects that target a specific quality attribute, compared to the ad hoc selection performed by development teams. Conclusions: The framework codifies a new risk-based understanding about how QATs impact product qualities across the development lifecycle. The selection method can be used to systematically choose the QATs for projects to target specific product qualities throughout lifecycle.