Simulation-based Internal Variable Range Coverage Metric and Test Generation Model

X. Cheng and M.S. Hsiao (USA)


software testing, partition testing, internal variable range coverage


In this paper, we present a new subdomain-based testing strategy which is associated with a new coverage metric called Internal Variable Range coverage. Unlike the traditional input domain partition technique, we apply partitioning on the value ranges of critical internal variables in a program. A simulation-based approach is proposed for range analysis and definition according to the variable data value distribution as well as the program structure information related to the internal variables. In addition, we exploit a sub-range transition finite state machine model to facilitate an internal variable range coverage oriented automatic test data generation process. Our experimental results on a class of extended triangle programs showed that our model can generate test set that achieves up to 44% higher internal variable range coverage than random test set as well as with better quality in terms of mutant killing, especially for applications that are more arithmetic computation related.

Important Links:

Go Back