Structural Test Data Generation Using an Alterable Genetic Algorithm for Large Scale Branch Coverage

Structural testing, also known as white-box testing, requires the generation of input data that ensures coverage of program structures such as statements, branches, and paths.The complexity of software development makes testing extremely challenging and demands novel approaches with significant advancement in the field.Manual testing by contrast, remains time-consuming and costly activity, accounting for almost 50% of software production costs.To address these challenges, several automated techniques have been explored, among which Genetic Algorithms (GAs) have emerged as one of the most widely adopted and studied approaches.GAs are a class of search-based optimization techniques inspired by natural selection, and they are particularly effective in solving complex search problems.However, the use of traditional GAs for structural test data generation suffers from premature convergence and population stagnation.To address these limitations, we introduce the Alterable Genetic Algorithm (AGA), a novel approach in which a single genetic operator is applied per iteration.An adaptive alternation function dynamically selects the most appropriate operator for the current state of search, capitalizing on their respective strengths to guide the search more effectively.This paper investigates the extent to which AGA improves structural test data generation, particularly by achieving high branch coverage with fewer fitness evaluations.