Experimental Analysis of m-ACO Technique for Regression Testing

Abstract

Listen

Translate article

Objectives: Experimental evaluation of “m-ACO” (Modified Ant Colony Optimization) technique for test case prioritization has been performed on two well known software testing problems namely “Triangle Classification Problem” and “Quadratic Equation Problem”. Apart from these two problems, m-ACO has been experimentally evaluated using open source software JFreeChart. Methods: m-ACO finds the optimized solution to test suite prioritization by modifying the phenomenon used by natural ants to reach to its food source and select the food. This paper attempts to experimentally and comparatively evaluate the proposed m-ACO technique for test case prioritization against some contemporary meta-heuristic techniques using two well known software testing problems and open source problem. Performance evaluation has been measured using two metrics namely APFD (Average Percentage of Faults Detected) and PTR (Percentage of Test Suite Required for Complete Fault Coverage). Findings: The proposed technique m-ACO proves its efficiency on both the parameters. m-ACO achieves higher fault detection rate with minimized test suite as comparative to other meta-heuristic techniques for test case prioritization. Improvements: The proposed technique m-ACO basically works by modifying the food source searching and selection pattern of the real ants. Real ants grab every type food source it comes across; while modified ants evaluate the food fitness and uniqueness before selection. This phenomenon enhances the quality and diversity of deposited food source. Keywords: Fault Coverage, Genetic Algorithm, Regression Testing, Software Testing, Test Suite Prioritization