Minimization Of Cost For Software Based Mutation Testing Using Surrogate Optimization Approach

Abstract

Mutation testing is the efficient and costlier testing method for the fault identification process for the code. It is costlier one due to the different levels of operations to generate the test case scenario. But, an application should pass such type of test cases to perform in an effective manner. Several approaches were proposed to reduce the cost for the mutation testing. The approaches were genetic algorithm, multi-objective particle swarm optimization and it is based on minimizing the cost for generating the test cases. But, the minimum value cannot be determined effectively because it is based on the bounds given to the problem and its search space region.  In this, the objective function for all the optimization approaches is same. Here, the cost function is based on the reachability, necessity and control parameters. The use of optimization approach is to determine the minimal value for the cost function with intensive searching which is similar to the mutation testing. The proposed multi-objective surrogate based optimization process to overcome the drawbacks of existing approaches. Because, the surrogate optimization main work is to determine the minimal value for the objective function. This property helps to reduce the cost effectively as compared to the previous approaches. Here, the cost function is based on the reachability, necessity and control parameters. The optimal value for this cost function will be determined through our proposed method and it will produce the best test case with minimal cost. The proposed method is implemented and tested using the emujava and it able to achieve high mutation score with lesser number of iterations.  It also able to identify the doubtful mutants which helps to speed up the process and reduce the test cases.