N-Gram based test sequence generation from finite state models

Abstract

Model based testing offers a powerful mechanism to test applications that change dynamically and continuously, for which only some limited black-box knowledge is available (this is typically the case of future internet applications). Models can be inferred from observations of real executions and test cases can be derived from models, according to various strategies (e.g., graph or random visits). The problem is that a relatively large proportion of the test cases obtained in this way might result to be non executable, because they involve infeasible paths. In this paper, we propose a novel test case derivation strategy, based on the computation of the -gram statistics. Event sequences are generated for which the subsequences of size respect the distribution of the -tuples observed in the execution traces. In this way, generated and observed sequences share the same context (up to length), hence increasing the likelihood for the generated ones of being actually executable. A consequence of the increased proportion of feasible test cases is that model coverage is also expected to increase. © Springer International Publishing Switzerland 2014.