The Application Of Evolutionary Testing To Nested Branch Structure Of Assembly Language In Embedded Software

Evolutionary testing is an emerging technology for automatic test data generation, which has been successfully applied in practice. It reformulates the test data generation problem into an evolutionary search under the guidance of a fitness function, a well-designed fitness function is crucial to the efficiency and effectiveness of evolutionary search. Therefore, the fitness function plays a very important role in evolutionary testing recently.In the application of evolutionary testing based on assembly language in embedded Software, the first problem faced is the program instrumentation, comparing to the high-level language, the assembly language have many different ways of addressing mode, therefore, the program instrumentation is much more difficult. This paper proposes an approach to solve this problem. Meanwhile, because the branch structure is the main structure of assembly language and the nested branch is the most complicate, therefore, we will focus on the fitness function calculation in the application of evolutionary testing based on nested branch structure of assembly language in embedded software. There are three methods in the calculation of fitness function, which are based on traditional method, the approximation level and the approximation level. The traditional method of fitness function design focuses on the approximation level, and can't effectively evaluate the data of internal branch, On the other hand, the method focus on the branch distances just concentrates the distance of branch and ignores the approximation level; but the method of fitness function based on the optimism level solves the problem above and makes the evolutionary search much more efficiency and effectiveness.Experimental results show that the fitness functions with the method based on optimism level can effectively improve effectiveness of evolution search.