ProMoSA – Probabilistic Model-based Safety Analysis

Fotolia_43886329_XSProject Members: Frank Ortmeier
Project Time: 01.01.2011 – 31.12.2013 
Funded By: Deutsche Forschungsgesellschaft/ German Research Foundation

Due to the ever increasing complexity and usage of software intensive systems in potentially safety-critical applications there is an increasing demand for sound safety analysis methods.

The goal of formal model-based safety analysis is to reliably derive safety properties inherent to the system from a model of the system and its environment. Through recent developments in computer science it became possible to not only determine qualitative measures, but also derive quantitative overall system failure probabilities from a model. The technology used for the approaches evaluated within the research project is based on stochastic models (Markov decision processes), verification techniques (symbolic and stochastic model checking) and multi-objective optimization (genetic algorithms).

The aim of this project is to provide safety measures to adjust the interaction between humans and technical systems in a way that ensures that neither humans nor their environment will be harmed. The technological foundation for these safety measures are the creation and formal analysis of formal abstract system models. As a system model cannot be analyzed meaningfully without its context, the system model is extended with an environmental model. Especially the environmental parts of the model may imply stochastic/nondeterministic elements to express human behavior (i.e. if and how often a driver ignores certain warnings) and physical effects (i.e. dynamics of the braking system of a car, or failure rates). Based on comprehensive models it is even possible to optimize a system to increase safety and/or performance. The main research challenges are to model the system and its environment appropriately, and to enable efficient automatic analysis of this model.

Last Modification: 10.11.2023 - Contact Person: Webmaster