A model independent S/W framework for search-based software testing

Area of Science:

• Software Engineering
• Software Testing
• Artificial Intelligence in Software Engineering

Background:

• Model-Based Testing (MBT) utilizes Search-Based Software Testing (SBST) for automated test case generation from system models.
• Current SBST approaches require reimplementation of search algorithms when model types change, leading to significant time and effort expenditure.
• This redundancy hinders the efficiency and scalability of SBST in diverse MBT applications.

Purpose of the Study:

• To propose a novel, model-independent software framework for SBST to address the challenge of model type variability.
• To reduce redundant development efforts and accelerate the application of SBST techniques across different model types.
• To demonstrate the effectiveness and efficiency of the proposed framework through empirical case studies.

Main Methods:

• Development of a reusable, common software platform designed to be independent of specific model types.
• Integration of design patterns within the framework to facilitate test case generation for various target models.
• Utilization of common functions provided by the framework to minimize development time and effort.
• Validation through two distinct case studies comparing the framework's performance against traditional approaches.

Main Results:

• The proposed model-independent framework successfully reduces redundant work in SBST.
• The framework provides a reusable platform that significantly decreases development time and effort.
• Case studies demonstrate a productivity improvement of approximately 50% when switching between different model types.

Conclusions:

• The developed model-independent SBST framework is effective and efficient for generating test cases across diverse model types.
• The framework offers a practical solution to overcome the limitations of model-specific SBST implementations.
• Adoption of this framework can lead to substantial gains in software testing productivity and resource optimization.