Applying Model Intelligence Frameworks for Deployment Problem in Real-Time and Embedded Systems

摘要

There are many application domains, such as distributed real-time and embedded (DRE) systems, where the domain constraints are so restrictive and the solution spaces so large that it is infeasible for modelers to produce correct solution manually using a conventional graphical model-based approach. In DRE systems the available resources, such as memory, CPU, and bandwidth, must be managed carefully to ensure a certain level of quality of service. This paper provides three contributions to simplify modeling of complex application domains: (1) we present our approach of combining model intelligence and domain-specific solvers with model-driven engineering (MDE) environments, (2) we show techniques for automatically guiding modelers to correct solutions and how to support the specification of large and complex systems using intelligent mechanisms to complete partially specified models, and (3) we present the results of applying an MDE tool that maps software components to Electronic Control Units (ECUs) using the typical automotive modeling and middleware infrastructure.