A Novel Algorithm for Attacking Path Explosion in Model-Based Test Generation for Data Flow Coverage

This paper presents a new efficient on-the-fly algorithm for test data generation to cover all feasible definition-use associations. It is well known that the ability to handle the exponential growth of the search space is the most critical feature of space exploration based methods. In this respect, the proposed algorithm stores and dynamically refines knowledge about sought coverage items reached from each state to prune the remaining exploration, so that a state will not be explored until it can increase the sought coverage, however, an early-terminated path can later be resumed upon the refinement in order to hold completeness of the search. While the asymptotic complexity is not improved and memory consumption is even increased, in many practical cases the algorithm terminates much earlier.