Information Technology Reference
In-Depth Information
From “Formal Methods” to System Modeling
Manfred Broy
Institut für Informatik, Technische Universität München
D-80290 München Germany
broy@in.tum.de
http://wwwbroy.informatik.tu-muenchen.de
Abstract.
When engineering software intensive systems the quality of the
resulting product depends strictly on the quality of the models used explicitly or
implicitly in the engineering process. A rich family of such models has been
developed in recent years. We discuss some of these models and describe the
requirements for system modeling theories.
1 Introduction
There is a long way from early approaches to formal system modeling sometimes
called “formal methods” such as denotational semantics, VDM, SADT, algebraic
specification to model based system and software engineering as advocated in wide
spread approaches such as UML, SysML, or MDA. We discuss this development and
properties of the theory and methodology that we require.
System and software development is today one of the most complex and powerful
tasks in engineering. Modern software systems typically are embedded in technical or
organizational processes, distributed, dynamic, and accessed concurrently by a variety
of independent user interfaces. Just by formulating the right programs we obtain
engineering artifacts that can calculate results, communicate messages, control
systems, and illustrate and animate all kinds of information. Since programs are -
implicitly or explicitly - based on models of system behavior and since well-chosen
models are a successful way to understand software, modeling is an essential and
crucial issue in software construction.
In all scientific and engineering disciplines, models play a prominent role. For
physics, mathematics has provided lots of models. The same holds for many
engineering disciplines. Economy works with models; biology works more and more
with models, chemistry works with models. Constructing, analyzing, and arguing in
terms of models is at the heart of science.
In informatics modeling is even more crucial. Developing software is more or less
nothing than developing the right models finally represented in the appropriate
notation such that they can be executed effectively and efficiently on today's
computing devices.
Many different models are needed in the engineering of software intensive
systems. To name a few of them:
•
Domain models: describing properties of the application domain that are
relevant for the system under development (physical, technical,
organizational, fiscal, legislation rules and laws),
Search WWH ::
Custom Search