One way of evaluating the robustness of a web service is to test it against invalid inputs. We introduce a model-based mutation technique which automatically generates faulty test inputs. From the specification of a Web service, a test model is designed using Uppaal Timed Automata and the conformance between the model and the implementation is validated via online model-based testing with the Uppaal Tron tool. A set of mutation operators is applied to the test model in order to generate mutant test models. We validate all generated mutants via verification rules and select those that are executable and introduce proper mutations. We employ bisimulation as a tool for detecting and eliminating equivalent mutants, that is those mutants which have identical input-output behavior with the original test model. The resulting mutants are used for online test generation against the service implementation in order to check whether the latter allows for unspecified behavior. We discuss tool support and present an experiment of applying our method for a case study of a blog web service with real-life properties. The experiment shows that the proposed approach of mutating the specifications is effective in detecting errors both in the system functionality and in the test model.
We present an evaluation study on applying aspect-oriented modeling concepts in UPPAAL timed automata. The study is focusing on the modeling and verification effort that can be reduced when applying explicit aspect-oriented structuring principles in model construction. We discuss the drawbacks and benefits related to model update and verification effort. The approach suggested is benchmarked on a mission critical crisis management system case study. We demonstrate the usability of our approach by extracting the aspects such as resource authentication and mission execution. Finally, we demonstrate by experimental data how our approach is more efficient compared to verification and testing effort applied in the non-aspect-oriented model.
As of today, model-based testing is considered as a leading-edge technology in the IT industry. In model-based testing, an implementation under test is tested for compliance with a model that describes the required behaviour of the implementation. Uppaal Tron is a popular tool for online model-based conformance testing of real-time systems; it uses the Uppaal verification engine to generate and convert on-the-fly timed symbolic traces into concrete test sequences. Among the advantages of online testing is the reduction of the symbolic state space needed for computing traces, better addressing non-determinism, as well as the possibility to execute longer-lasting test runs. However, analysing and debugging long test runs can be tedious and time-consuming especially when analysing root causes of failed tests. In game theory, backward-induction is a process to reason backwards in time, from the end of a problem or situation, in order to determine a sequence of optimal actions. In this paper, we propose an approach to reconstruct symbolic traces from test sequences generated by Uppaal Tron using backward-induction. The resulting symbolic traces can be imported in the Uppaal tool and visualised in the Uppaal simulator. The evaluation of the implementation of the approach shows that it has the potential to satisfy the needs of industrial level testing.