Designing And Researching The Method To Produce Test Case Set Of K-n Software System

Pair-wise testing is a practical and effective testing for various types of software systems. At present, there are mainly four methods to construct test case set such as OAM(Orthogonal Array Method),CAM(Covering Array Method),IPO,AETG and so on.In this thesis, first of all we introduce the main thoughts of these four kinds of methods. We also summarized and compared their own ranges of application and both their advantages and shortcomings. Then we analyze the course of constructing test case set by Orthogonal Array Method and Covering Array Method, and their application to the pair-wise testing. We analyze the problems when we use these methods in the software testing to generate test case set, including no adaptability between the software testing and the method to generate test case set, complicated algorithm to construct test case set and much more test cases and so on, so we put forward the improved methods.In order to solve a great deal of problems existing in these methods effectively, we give a k-n software testing, which means we divide test systems into four types according to numbers of parameter k and numbers of their values n. As for two-value software system, this paper designs an algorithm to construct test cases, based on the longitudinal binary string set and method of controlling variables. As for k-n software system which n is larger than 2 and n is prime or power of prime, we design a method of covering array of test set to construct test set by taking orthogonal array and deriving arrays from orthogonal array and two useful conditions. As for a software system with k parameters whose n-value is not all equal, we design an experience algorithm.The thesis verifies the validity of these three kinds of methods with the experiments finally. The experimental results show that the size of test sets from the methods we give in this paper is superior to the size from OAM,CAM,IPO, AETG. And the adaptability between the software testing and the method to generate test case set has been improved greatly.