Research On Particle Swarm Optimization-Based Test Cases Generation And Prioritization

With the advance of modern information society, software system has become animportant part of people’s daily lives. Software error may be related to issues ofpersonal and property safety, and even cause huge economic losses. Therefore, toensure the quality of the software system and improve the efficiency of softwaretesting, more advanced software testing technology is needed. As the key link ofautomated software testing, the way of software test cases automatic generation cangreatly reduce the cost of software testing and improve the efficiency of softwaretesting, and test cases prioritization can further improve the cost-effective ratio ofsoftware testing. As we all know, search algorithm is considered to be the mostpotential way for test case automatic generation and prioritization. Although there arelots of domestic and foreign researches, it is still not mature enough. So it is worth toresearch on the automatic test cases generation and prioritization based on the searchalgorithm.This paper mainly researches on test cases automatic generation andprioritization based on particle swarm optimization. For the problem of test casesgeneration, the framework of automatic generation of test cases based on PSO ispresented, which is an effective solution for further study. To verify the validity ofPSO for test cases generation, this paper makes a comparative study on PSO with GAand SA from four aspects which are the average branch coverage, the average successrate, the average convergence generation and the average convergence timerespectively. Study indicates that PSO is faster and more efficient. At the same time,contrast analysis is described between PSO and its four typical variants for test casesgeneration. For the problem of test cases prioritization, an improved PSO-based testcases prioritization way is presented. Experimental results show that it is significantlyhigher than the greedy algorithm, and equivalent with additional greedy algorithm.Therefore, the ways presented in this paper, which are simple, fast and effect, provideefficient solutions for test cases generation and prioritization problems.