The increased use and interconnection of electronic components in automobiles has made communication behavior in automotive networks drastically more complex. Both communication designs at application level and complex communication scenarios are often under-specified or out of scope of existing analysis techniques. We extend traditional protocol analyzers in order to capture communication at the level of abstraction that reflects application design and show that the same technique can be used to specify, monitor and test complex scenarios. We present CFR (Channel Filter Rule) models, a novel approach for the specification of analyzers and a domain-specific language that implements this approach. From CFR models, we can fully generate powerful analyzers that extract design intentions, abstract protocol layers and even complex scenarios from low level communication data. We show that three basic concepts (channels, filters and rules) are sufficient to build such powerful analyzers and identify possible areas of application.

Abstract. Today, component-based software and system development has gained considerable attention and is wide-spread. Components and/or modules manifest the well-proven engineering principle of divide and conquer. Formal approaches have helped to provide a theoretical foundation to component-orientation. There is a broad range of system architectures in practice today in terms of layered distributed open systems (as for example known in telecommunications). However, the mathematical formalisms developed so far are rarely applied to these practical architectures. In fact, the notion of distribution and stacked communication layers is hardly understood precisely and not supported by architecture description languages. This article addresses this gap. The formal conditions of distribution and layering are investigated and a mathematical model for layered distributed systems is presented. Communication refinement and so-called complex connectors turn out to be vital concepts which are not addressed in today’s modeling languages. In addition to that, two important design approaches are derived for the architectural design of complex communication systems: A node-centric and a network-centric design approach. Both can significantly improve the design process.

Do we have the right set of concepts for modeling and describing distributed real-time systems in software engineering? Languages like ROOM (Real-time Object-Oriented Modeling) and UML-RT (Unified Modeling Language for Real-Time) offer some basic concepts; however, system engineers of the application domain (e.g. in telecommunications) work with a variety of concepts which do not have a corresponding counterpart in software engineering. There seems to be some potential for transfer. This report explores the power of a well-known concept in telecommunications, the service access point (SAP). SAPs not only prove to be more powerful than layering is, but they even subsume the concept of planes, which are almost unknown in software engineering. SAPs turn out to be a basic modeling construct, which allows us to describe in a precise manner any communication architecture that is structured in layers and planes. In this paper, it is shown that a few improvements to ROOM/UML-RT can make the full power of SAPs available to the object-oriented design of distributed real-time systems.