Modelling of autonomous components for heterogeneous intelligent environments (MACHINE)

(Internally funded)

Motivation

Currently, we experience a strong interest in exploring the modeling and realization of applications which we may call traditional in the sense that these applications show interactive behavior, keeping immediate contact to the user, and being restricted to only a few (or maybe only one) network node. These are well-known applications or tools like editors, media players, or information exchange tools.

Research in the area of information technology shows that we will encounter an increasing number of ubiquitous systems with execution capabilities. While formerly we had only one computer for a number of users, the situation seems to turn around now. The single users becomes more and more overloaded with management tasks which take away valuable time for working on the actual problems.

In this context, the term Ubiquitous Computing keeps appearing in the discussion. Systems merge with the environment; we do not find them below our desks anymore. Users interact with these systems in many ways, and vice versa, these systems react to the presence of users and objects.

On one hand, we require concepts for maintaining those systems in a reasonable and effective way. This may be achieved by some intelligent support of the user or by creating systems that employ so-called self-properties. On the other hand, novel sofwtare development processes need to be conceived to fully exploit the potential of these technologies.

Traditional software engineering has gained considerable interest in the area of model-driven approaches (like MDD). Techniques like the utilization of UML enable the designer to create application models of various degrees of abstraction, and thus allow the cooperation between domain and platform experts.

However, despite the broad acceptance of MDD, it cannot give an answer how systems with distributed control can be effectively and efficiently created. This is mainly due to the traditional view on computer systems as interactive systems, woven into the practise of software design.

Project objectives

It is our goal to provide full support for modeling systems composed of autonomous components. We utilize results from the AgentUML research and further agent-oriented modelling techniques in order to find

• novel design paradigms
• appropriate tools, and
• illustrative scenarios.

During the whole project lifetime, results will be disseminated as prototypes and research papers at specific workshops and conferences.

For further information contact Dr. Michael Zapf (former group member).