Towards Model-Based Development of Self-Managing Automotive Systems : Modeling, Simulation, Model Transformations and Algorithms: Supporting the Development of the DySCAS Middleware

Abstract: With the increased usage of electronics, embedded computer systems, software and networks in modern vehicles, the complexity is also increasing in terms of the development process, functionalities as well as their implementation. On one hand there is a need for efficient and systematic development, and on the other hand the systems are required to have enhanced post-deployment time configuration management support. This promises not only to reduce the time to market but also increase the development and maintenance efficiencies. Moreover, the possibility to adapt system behavior and structure by software can improve the overall traffic infrastructure, increase robustness and safety and at the same time enable changeable user defined configurations and addition of new features throughout a vehicle's life cycle. However, the dynamic configuration management characteristics increase software complexity and provide new failure modes. Therefore, there is a need for appropriate development support. This thesis addresses model-based development and dynamic configuration management for automotive systems in context of the DySCAS project. The framework for dynamically self-configuring middleware from the DySCAS project is the main object of study. The major focus is on algorithms for dynamic reconfiguration, model transformation, modeling and simulation. Requirements for self-managing systems with self-configuration, self-healing, self-optimization and self-protection characteristics and the simulation platforms for early analysis, verification and validation are also identified. A mapping scheme from UML™ to SimEvents™, evaluation of tools for simulating self-managing systems, simulations of dynamic reconfigurations and a foundation for a complete simulation environment are the major outcomes of this thesis.