In this paper, we present a general framework for testing time-critical systems and software, as it is proposed in the European IST project MaTeLo. The main focus is on automatically generating a MCUM (Markov Chain Usage Model) starting from an FDT (Formal Description Technique) description in order to derive TTCN-3 (Testing and Test Control Notation version 3) compatible test case definitions.Our approach is a combination of statistical usage testing based on a given MCUM and specification-based testing that is using FDT inputs. Within MaTeLo, special attention is given to international standardized FDT notations, specifically ITU-T MSC (Message Sequence Chart). In addition, we make use of annotations to specify selected non-functional requirements to support the automated software testing of the real time systems. We also defined an XML-based representation format called MCML (Markov Chain Markup Language) to build a common interface between various parts of the MaTeLo tool set.

No abstract

We describe model-level unit tests for model-driven simulation based on UML models. We refer to the well-known unit testing method and apply this concept on a higher abstraction level, that is on UML simulation models. The concept of model-based simulation has become more and more popular throughout the last years. This trend is fostered by the availability of tools that automatically transform UML models into executable simulation code. Typically, both functionality and behavior are modeled in UML, whereas debugging and validation are mainly an issue of investigating the executable code. We contribute to the field of model-driven simulation by defining a novel testing method. Our method allows to use UML to specify model-level unit tests in order to validate simulation models defined with UML. In addition, we describe the translation, execution and evaluation of these tests within the framework Syntony. In this paper, we show the principles of this method and discuss its use in the scope of our simulation framework Syntony as well as its general applicability. Our implementation allows to compile and to execute the test code in combination with the simulation code. In spite of the high abstraction level, the full functionality of unit testing is provided and the modeler can rely on automated test case generation and execution. After execution of the tests, the achieved test coverage is computed as a measure for the test quality.

No abstract

We will first give a brief introduction to a versatile modeling environment that allows the early validation of system specifications by a test-driven agile simulation approach. The main focus of our paper is on providing techniques for automated test case generation relying on statistical usage models. Based on the open source software R for statistical computing and graphics the easy to handle test case generation and analyzing TestPlayer c tool was developped. Starting from customer-specific graphical Markov chain usage models test cases are automatically generated and visualized by highlighting selected nodes and arcs in the model. In addition, various metrics and corresponding diagrams offer analytical techniques to assess the quality of the derived test suite.