Article Preview
TopIntroduction
The automated test data generation is an undecidable problem and the main challenge in this field is the size and complexity of the software. Nature inspired algorithms are good at seeking good solutions for combinatorial optimization problems within confined time at a reasonable computational cost. A good number of Nature based algorithms coming under metaheuristic have been used by researchers to generate test data for white box, black box as well as grey box testing (Harman, 2015). Most of the research work has been done for structural testing, functional testing, mutation testing etc., but very few papers are available related to grey box and quality testing (Harman, 2009). The first workshop on software engineering using metaheuristic algorithms was conducted in 2001 (Harman, 2001). At that time only limited work has been done by the researchers in the area of software testing using search algorithms like Simulated Annealing (SA), Tabu search and Genetic algorithms (Harman, 2001; McMinn, 2004; Sthamer, 2002).
Search based software engineering (SBSE) is a reformulation of the software engineering problems (Harman, 2004; de Oliveira, 2015) coming under the category of un-decidable, NP hard or NP complete problems. Here the specific problem is designed as a search problem and then the target is to select best solutions out of a large domain of solution space using guided heuristics mimicked from nature (Harman, 2015).
Search based techniques are perfect for the software engineering problems as this is the area where the engineers have no idea regarding what are the good solutions but very clear idea about what qualities the good solutions must have, and they have to choose the best one out of a number of competing and conflicting solutions(Harman, 2001; McMinn, 2004).To redesign any problem of software engineering as a search problem, three components are necessary to be specified clearly, these include first an intermediate representation of the problem, second a fitness function on the basis of the targeted objective and finally a set of operators to deploy (Harman, 2001).
Out of these three the most crucial role is played by the fitness function. The fitness function captures the vital information and guides the search process by selecting the best solutions from a number of suitable solutions on the basis of the testing strategy adopted (Sthamer, 2002; Baresel, 2002). For example, in white box testing if path based testing approach is followed then the suitable test cases are those test cases that are able to show maximum path coverage.