Goal-Based Domain Modeling as a Basis for Cross-Disciplinary Systems Engineering - Intentional Perspectives on Information Systems Engineering - page 86

Requirements
Analysis

Acceptance
Test

Control
Engineers

Architecture
Design

Integration
Test

Module/ Algo-
rithm Design

Module
Test

Software
Engineers

Implemen-
tation

Fig. 1 Development process from control system engineering point of view [22]

development, the model of the controlled system is at the center of interest, in
particular to enable simulations as described above. Furthermore at the level of
requirements, textual approaches still prevail [ 15] . In contrast, models within soft-
ware engineering focus the system to be developed and put these models at the
center of the whole development process as envisioned in the model-driven archi-
tecture (MDA) promoted by the OMG [ 29] . Domain-specific modeling languages
(e.g. [ 2, 12] ) are intended to be easier to use for domain experts; recent empirical
evidence [ 8] demonstrates that this is indeed the case. Even the Matlab/Simulink
environment, most commonly used by control system engineers, can be consid-
ered a domain-specific language, but only for the design and implementation level.
Besides architectural models, model-based requirement specifications enable a bet-
ter structuring, traceability and a smarter transition from requirements to subsequent
development steps [ 44] .

2.2 SMEs Developing Control Systems

In the control systems development sector, SMEs play an important role as inno-
vation drivers that perform engineering tasks for multiple customers. The process
is typically initiated by an engine manufacturer (building engines for cars, boats,
power saws, or the like) contacting a number of suppliers to provide an offer to
develop a control system for a new engine. By nature, the time frame for the sup-
plier to respond is very short. The manufacturer delivers a specification of the engine
and the list of required control functionality. To prepare the offer, the supplier first
specifies the requirements on the requested control system from a developer's point
of view. In a second step, he prepares a system design in order to calculate the costs
for the development of the control system. To keep the development costs low and
to win the contract, he must on the one hand reuse as many software artifacts and
simulation models as possible from previously developed control systems. On the