Adaptive random testing with randomly translated failure region

  • Authors:

  • Johannes Mayer

  • Affiliations:

  • Ulm University, Ulm, Germany

  • Venue:
  • Proceedings of the 1st international workshop on Random testing
  • Year:
  • 2006

Quantified Score

Hi-index 0.00

Visualization

Abstract

Adaptive Random Testing (ART) algorithms are designed to be more effective than Random Testing. Some of these methods however distribute the test cases not evenly within the input domain. Therefore, some locations are preferred. Since the locations of failure-causing inputs of a system under test are obviously unkown, such a preference makes the method more effective for some systems under test and less effective for others. This paper addresses the described problem and tries to equalize the effectiveness of a testing method for all systems under test whose failure-causing inputs have the same geometric shape. Virtually, all failure-causing inputs are randomly translated to reach this goal. This method is applied to two well-known ART methods that tend to generate test cases at the corners and the boundary of the input domain more frequently. However, the presented method is not restricted to any strategy for test case selection.