Parameter Adjustment of Test Profile in Dynamic Random Testing

Abstract

The basic motivation of Dynamic Random Testing (DRT), which is proposed in the context of software cybernetics, is to use test results as the feedback information to realize dynamic adjusting of selection probabilities of all the test cases. It is unreasonable that the profile adjustment in the original DRT is somehow contingent and only utilizes the latest execution result. Moreover, the parameter adjustment of test profile for different subdomains is uniform in original DRT, which may not the best solution to improve the effectiveness of DRT. This paper presents an improved dynamic random testing that based on the function information of test cases (DRT-F) which makes use of correlation among the test cases to improve the performance of the original DRT. In DRT-F, the parameter adjustment is according to the function information of test case classification and makes the DRT adaptive to different classification of its influence on its performance.