Hybrid adaptive random testing

Abstract

Adaptive random testing (ART) subsumes a family of random testing techniques with an effective improvement. It is based on the observation that failure causing inputs tend to be clustered together. Hence the ART methods spread test cases more evenly within the input domain to improve the fault-detection capability of random testing. There have been several implementations of ART based on different intuitions and principles with their own advantages and disadvantages. In the different variants of ART methods, the majority of them use a variety of distance calculations, with corresponding computational overhead. The newly methods try to decrease computational overhead while maintaining the performance through partitioning the input domain. We outline a new partitioning-based ART algorithm with a hybrid search method and demonstrate experimentally that it can further improve the performance, with considerably lower overhead than other ART algorithms.