Automatic Test Data Generation For Path Testing Using Ant Algorithms

With the development of informationzation, software products being used more and morewidely, the system getting more and larger and complexity, the quality of software becomes ahotspot and difficulty problem of computer technical field. Software testing is a very importantmeasure to assure quality and reliability of software. Many problems in testing such asstatement coverage and conditions coverage, can be reduce to the pathwisetest data generatingproblem.This paper focuses on automatic test data generation for path testing using ant algorithms.Path testing is a problem of NP, and also a main method of structural testing. Unfortunatelymost software testing works are finished by people. In fact, however, many operations in wokof testing are repeated, which support probability to testing using computer. Studies ofautomatic test data generation can reduce the high cost of manual software testing and at thesame time increase its reliability, especially research automatic test data generation for pathtesting, which is the fundamental of entire test cycle.First,this paper introduces several steps of test data generator system. Main contents ofevery step and which one in use is also given out in this dissertation. After analysis some mainway of automatic test such as random algorithm, static and dynamic method, those are maintechnique of automatic test. As one of the most successful swarm intelligence algorithm, antcolony optimization (ACO) solve almost NP problem, but few in path testing. The goal of thispaper is solve automatic test data generation using three binary ant algorithms.Catch tightly, this paper introduces the basic principle of ant algorithm. As result of thecharacteristic of ant solve a problem, which is based on search space. The first problem when using ant algorithm to resolve path testing is how to constructs the search space, which is binarysearch space in this paper, that the ants can search the solution. After that, this dessertation usesBAS, BAA and BACS to search the best solution and analysis it.For the specialty of binary search space, there is no heuristic in binary ant algorithm. Sothe local search is very important. After analysis the efficiency of three algorithms, then addlocal search into the best one. During the search, for balance the exploration and exploitation,we can tune some parameters, avoid algorithm trap in prematurity.At last, as an example, we generate testing data for the Program of Triangle Classifier andCommission Problem. The result shows BAS with local search have nice effect in automatictest data generation and assure that the diversity of solutions. So with the experiment, thevalidity of Binary Ant Algorithm can be clearly seen.