Adaptive random testing through iterative partitioning

Random testing (RT) is a fundamental and important software testing technique. Based on the observation that failure-causing inputs tend to be clustered together in the input domain, the approach of Adaptive Random Testing (ART) has been proposed to improve the fault-detection capability of RT. ART employs the location information of previously executed test cases to enforce an even spread of random test cases over the entire input domain. There have been several implementations (algorithms) of ART based on different intuitions and principles. Due to the nature of the principles adopted, these implementations have their own advantages and disadvantages. The majority of them require intensive computations to ensure the generation of evenly spread test cases, and hence incur high overhead. In this paper, we propose the notion of iterative partitioning to reduce the amount of the computation while retaining a high fault-detection capability. As a result, the cost effectiveness of ART has been improved.