A multiple-population genetic algorithm for branch coverage test data generation

Abstract

The software testing phase in the software development process is considered a time-consuming process. In order to reduce the overall development cost, automatic test data generation techniques based on genetic algorithms have been widely applied. This research explores a new approach for using genetic algorithms as test data generators to execute all the branches in a program. In the literature, existing approaches for test data generation using genetic algorithms are mainly focused on maintaining a single-population of candidate tests, where the computation of the fitness function for a particular target branch is based on the closeness of the input execution path to the control dependency condition of that branch. The new approach utilizes acyclic predicate paths of the program's control flow graph containing the target branch as goals of separate search processes using distinct island populations. The advantages of the suggested approach is its ability to explore a greater variety of execution paths, and in certain conditions, increasing the search effectiveness. When applied to a collection of programs with a moderate number of branches, it has been shown experimentally that the proposed multiple-population algorithm outperforms the single-population algorithm significantly in terms of the number of executions, execution time, time improvement, and search effectiveness.

Applying Genetic Algorithms to Test JUH DBs Exceptions

“…'); [22] elsifp_error_no = 5 then [23] raise_application_error ( -20006, 'Pricing datais incompleteforthis patient!! '); [24] elsifp_error_no = 11 then [25] raise_application_error( -20001, '11'); [26] elsifp_error_no = 21 then [27] raise_application_error ( -20003, '21'); [28] elsifp_error_no = 31 then [29] raise_application_error ( -20004 …”

Using Genetic Algorithm as Test Data Generator for Stored PL/SQL Program Units

“…It is an activity that helps finding out bugs and errors in a software system that is under development in order to provide a bug free and reliable system/solution to the customer [2]. Testing has two main types based on the knowledge of the system: black box testing (functional) and white box testing (structural) [2][3][4].…”

“…It is an activity that helps finding out bugs and errors in a software system that is under development in order to provide a bug free and reliable system/solution to the customer [2]. Testing has two main types based on the knowledge of the system: black box testing (functional) and white box testing (structural) [2][3][4]. The functional testing deals with the system as a black-box that does not explicitly use knowledge of the internal structure; which means it usually makes sure that the system is working according to the system requirements, while the structural testing generates the test data depend on the knowledge of internal code of the system.…”

“…During structural testing, the goal is to generate a test data which satisfy a given testing criterion to cover given elements of the program. In this paper, the branch coverage criterion is considered, where each branch of the program should be reached by some test data [2].…”

“…In contrast, dynamic testing techniques execute the program under test on test input data and then simply observe the results. Consequently, dynamic testing reduces the cost of software development and maintenance [2]. Search-based software testing is an example of dynamic method used to generate test set that can be successfully applied in structural testing.…”