Enhancing Path-Oriented Test Data Generation Using Adaptive Random Testing Techniques

摘要

In this paper, we have developed an approach to generate test data for pathcoverage based testing. The main challenge of this kind testing lies in itsability to build efficiently such a test suite in order to minimize the numberof rejects. We address this problem with a novel divide-and-conquer approachbased on adaptive random testing strategy. Our approach takes as input theconstraints of an executable path and computes a tight over-approximation oftheir associated sub-domain by using a dynamic domain partitioning approach. Weimplemented this approach and got experimental results that show the practicalbenefits compared to existing approaches. Our method generates less invalidinputs and is capable of obtaining the sub-domain of many complex constraints.