Optimal Testing Resource Allocation for modular software systems based-on multi-objective evolutionary algorithms with effective local search strategy

Software testing is a very important part in software projects. As a key issue in software testing, Optimal Testing Resource Allocation Problems (OTRAPs) have drawn more and more attention recently. Along with the rapid increasing of the scale and complexity of software systems, the problems become more and more difficult to solve. Although some single objective optimization approaches had been used to solve such problems, quite a number of flaws were observed with these approaches, such as trapping into local optima, high computational complexity and few available optimal solutions. In this paper, to solve the problem of few available optimal solutions, an effective local search (ELS) is introduced into two effective multi-objective evolutionary algorithms: Non-dominated Sorting Genetic Algorithm II (NSGA-II) and Harmonic Distance Based Multi-objective Evolutionary Algorithm (HaD-MOEA), advantages of this strategy over pure multi-objective approaches are testified on two OTRAPs with parallel-series modular software systems. To deal with the problem of high computational complexity, the proposed ELS is also embedded into another effective multi-objective algorithm, Multi-objective Evolutionary Algorithm based on Decomposition (MOEA/D) to solve OTRAPs. Comprehensive experimental studies show the better performance over the state-of-the-art multi-objective approaches for OTRAPs.