Characterizing failure-causing parameter interactions by adaptive testing

Combinatorial testing is a widely used black-box testing technique, which is used to detect failures caused by parameter interactions (we call them faulty interactions). Traditional combinatorial testing techniques provide fault detection, but most of them provide weak fault diagnosis. In this paper, we propose a new fault characterization method called faulty interaction characterization (FIC) and its binary search alternative FIC_BS to locate one failure-causing interaction in a single failing test case. In addition, we provide a tradeoff strategy of locating multiple faulty interactions in one test case. Our methods are based on adaptive black-box testing, in which test cases are generated based on outcomes of previous tests. For locating a t-way faulty interaction, the number of test cases used is at most k (for FIC) or t(dlog2 k] + 1) + 1 (for FIC_BS), where k is the number of parameters. Simulation experiments show that our method needs smaller number of adaptive test cases than most existing methods for locating randomly-generated faulty interactions. Yet it has stronger or equivalent ability of locating faulty interactions.