Logo des Repositoriums

Auflistung nach Autor:in "Kloos, Johannes"

1 - 2 von 2
  • Konferenzbeitrag
    Indicator-based inspections: A risk-oriented quality assurance approach for dependable systems
    (Software Engineering 2010, 2010) Elberzhager, Frank; Eschbach, Robert; Kloos, Johannes
    We are surrounded by ever more dependable systems, such us driving assistance systems from the automotive domain or life-supporting systems from the medical domain. Due to their increasing complexity, not only the development of but also the quality assurance for such systems are becoming increasingly difficult. They may cause various degrees of harm to their environment. Hence, in order to reduce risks associated with these systems, development as well as quality assurance normally use risk analysis as a basis for constructive and analytical measures against these risks. One of the aims of quality assurance is fault detection and fault forecasting. In this paper, the authors present indicator-based inspections using Goal Indicator Trees, a novel risk-oriented quality assurance approach for fault detection. It can be used to detect faults of different types, like safety faults or security faults. Starting from typical risk analysis results like FMECA and FTA, the approach systematically derives quality goals and refines these goals into concrete quality indicators that guide the indicator-based inspection. Quality indicators can be mapped to concrete checklists and concrete inspection goals in order to support inspectors checking artifacts in a fine-grained way with respect to certain quality properties. The approach is explained and demonstrated with respect to the quality property safety, but tends to be generalizable to further quality properties.
  • Zeitschriftenartikel
    Risikobasiertes statistisches Testen
    (Softwaretechnik-Trends Band 29, Heft 4, 2009) Zimmermann, Fabian; Eschbach, Robert; Kloos, Johannes; Bauer, Thomas
    In dieser Arbeit stellen wir erste Ideen für eine Methode zur automatischen Ableitung risikoreicher Testfälle vor. Diese Testfälle werden aus Modellen abgeleitet, die speziell zum Testen erstellt wurden. Das hier vorgestellte Verfahren ist eine Anpassung des modellbasierten statistischen Testens für risikoreiche Systeme. Dabei verwenden wir Markov-Ketten, die das Nutzungsverhalten beschreiben. Diese Markov-Ketten werden so verändert, dass nur risikoreiche Testfälle, die eine realistische Nutzung des Systems darstellen, erzeugt werden. Dadurch kann die Zuverlässigkeit des Systems in kritischen Situationen ermittelt werden.