The Research Of Test-suite Reduction Technology Based On Improved Ant Colony Algorithm

In recent years, with the constant expansion of the software, the growing complex ofsoftware development process, and due to the limit of people's subjective understanding,makes inevitably introduce errors and vulnerabilities in the software development process.These software errors and vulnerabilities may cause the entire software system failure, andcause significant economic losses, if the software is running in some key areas, such asnuclear power, may endanger the security of human lift. So, as the important measure toassure the quality and reliability of software, software testing has become a very importantpart of the software development process. During software testing, the design of software testplan and test case is a crucial part. In software testing, we must pay much attention to thesetwo aspects of the design. With the expansion of software size, the software requirements aremore and more, the corresponding number of test cases will be more and more, whichcontains a large number of redundant test cases. The number of test cases and the running costof each test case determine the cost and efficiency of software testing. Therefore, how toensure that adequate testing and software quality under the premise, it would cost less and lesspossible test cases to fully test the software, thereby reducing the cost of software testing,improve the efficiency of software testing is the focus of this paper.This paper first introduces and analyzes the existing several test suite reduction algorithms,and point out their own shortcomings, then provide an algorithm of test suite reduction basedon the initial pheromone enhancements, and applied to the problem of test suite reduction.Test suite optimization method proposed in this paper take full account of the relationshipbetween the test cases, and test suite optimization from the perspective of the test suitereduction. First, using GRE algorithm ideological to select the essential test case and save itfrom original test cases, and remove redundant test cases at the same time, reduct the originaltest case collection by recycling essential strategy and redundancy strategy. After pretreatmentof the test suite-the test requirements set relationship matrix, using the greedy algorithm togenerate an a prior test case subset, and enhance the initial pheromone of all nodes in subset,run the ant colony algorithm based on demand the optimal solution. A prior collection ofnodes in the initial pheromone enhanced to help reduce the randomness of the ants search, make algorithm attenuation from a good basis; meanwhile, the algorithm iterations of theadaptive changes shorten the running time, compared to a fixed number of iterations. Thisimproved ant colony algorithm can find a balance between in the convergence speed andoptimal solution. Finally, this paper designs a comparison tool to compare the improved antcolony algorithm with several classic algorithms, and display the comparison results in avisual form. The experimental results show that the new algorithm in finding the minimumcost of test case and to accelerate the convergence of the process performance is superior toother algorithms.