An Innovative Approach to Tackling the Boundary Effect in Adaptive Random Testing

Adaptive random testing (ART) is an effective improvement of random testing (RT) in the sense that fewer test cases are needed to detect the first failure. It is based on the observation that failure-causing inputs are normally clustered in one or more contiguous regions in the input domain. Hence, it has been proposed that test case generation should refer to the locations of successful test cases (those that do not reveal failures) to ensure that all test cases are far apart and evenly spread in the input domain. Distance-based ART and restricted random testing are the first two previous attempts. However, test cases generated by these attempts are far apart but not necessarily evenly spread, since more test cases are generated near the boundary of the input domain. This paper analyzes the cause of this phenomenon and proposes an enhanced implementation based on the concept of virtual images of the successful test cases. The results of simulations show that the test cases generated by our enhanced implementation are not only far apart but also evenly spread in the input domain. Furthermore, the fault detection capability of ART for high-dimensional input domains is also enhanced