Model And Evolutionary Solution For Test Data Generation Of Path Coverage Based On Variable Grouping

Computer software as an important component of the information industry.Ensuring the security, stability and reliability of the products has been a hot research issues and concerns in academia and industry. One of the important and indispensable measures to improve software quality is that require a lot of tests before putting in market using to detect a variety of defects what may exist in the software, and also need to make timely and correct repair and improvement defects. It is necessary to use of targeting theoretical approaches and generating high-quality test data to comply with the established test criteria when we test the software and keep its validity and reliability.It is a very efficient and classical testing method which is based on path coverage. Path coverage testing need to generate test data to cover all the target paths.However, some target paths are difficult to through for some complex software, and it is inefficient use of traditional methods to generate test data. Research achievements of generate test data based on path coverage that use evolutionary optimization approach have been putted forward. However, when the program contains numbers of variables, genetic algorithms is difficult to generate test data effectively, because the number of variables leads to a sharp increase in the search space issues and make the search much more difficult. In view of this, this paper presents test data generation methods that based on variable group which effectively improve the efficiency of test data generation.This paper researched the method of grouping variables, the variables are grouped effectively. Firstly, combined with the basic concept of Graph and Program Control Flow Graph, which are used to analysis relationship between the variables with node, program paths and variables, we also reference the correlation definition of variables. Secondly, we would like to translate high-dimensional multi-objective optimization problem into a plurality of low-dimensional sub-optimization problem according to the correlation between variables, and each sub-optimization includes a little of variables on the correspond target path, Lastly, Building mathematical model of the problem according to the result of variables grouping and evolutionary use of Co-evolutionary Genetic Algorithm for solving path coverage generated test data.Experimental result shows that the grouping strategy can improve the efficiency of test data generation.Secondly, this paper studied the method of the test data generation for multi-paths coverage based on paths grouping. First of all, Based on the idea of linkage of identification and divide variables into several groups; then according to the result of grouping variables, and combined with location information of variables in the nodes and the relationship between nodes with paths, consider the similarity of among paths to group the target paths of programs, So, we transformed the coverage problem of all the target path into a finite grouped coverage problem Next, we use the co-evolutionary algorithm for solving the model to optimizing the solving process and generates test data effectively. The experimental result proved the effectiveness of the method.Results of this research propose a new solution to solve test problem of the complex software which contain multi-variables and multi-paths, and improve the quality and efficiency of the test data generated. Therefore, it not only provides technical support for the protection of software quality, and also has important theoretical significance and practical application value.