Search-Based Software Testing Driven by Bioinspired Algorithms: A Systematic Literature Mapping

Bioinspired algorithms have gained relevance in Search-Based Software Testing (SBST) mainly because they can find multiple solutions to difficult-to-solve problems. Exploring the usage trends and elements of its implementation is necessary for software engineering since it allows for understanding the area's current state and proposing guidelines for innovation. We proposed a Systematic Literature Mapping to analyze the canonical bioinspired algorithms used in SBST. During the mapping conduction, 60 studies were analyzed, in which bio-inspired algorithms were used in different software testing activities, such as test data generation, test case generation, test case prioritization, and others. The practical implications of our research are significant. The most commonly applied bioinspired algorithms in these activities are the Genetic Algorithm, Particle Swarm Optimization, and Ant Colony Optimization. Concerning problem types, most of these algorithms are focused on single-objective, with or without constraints, and, to a lesser extent, multi-objective and phased problems. According to the results, SBST driven by bio-inspired algorithms favors variables such as execution time and computational cost. Code coverage is usually used as a performance metric to evaluate its effectiveness. The application of canonical bioinspired algorithms mainly seeks to solve single-objective problems.