23.08.2013 Views

Specification of Reactive Hardware/Software Systems - Electronic ...

Specification of Reactive Hardware/Software Systems - Electronic ...

Specification of Reactive Hardware/Software Systems - Electronic ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

1.1 Objectives 3<br />

understood and still a subject for research. For this reason we did spend a lot <strong>of</strong> research<br />

in defining a practically useful object approach.<br />

For a complex product the knowledge <strong>of</strong> technology is, although necessary, not the key<br />

to a complete product idea. A bottleneck today is the definition <strong>of</strong> what is to be designed.<br />

The required functionality <strong>of</strong> a system cannot be known before one has worked on a<br />

model <strong>of</strong> the system for a substantial amount <strong>of</strong> time. A specification in the form <strong>of</strong><br />

a traditional requirements list is absolutely inappropriate for complex systems. This<br />

approach assumes that one is able to determine in advance the requirements <strong>of</strong> the<br />

system to be developed. The traditional bottom up approach <strong>of</strong> designers does not lead<br />

to efficient product development anymore. The same holds for the traditional top down<br />

approach.<br />

Emphasising technology leads to the well known pitfall <strong>of</strong> ’technology push’. An<br />

alternative is a market-driven approach. The premise is <strong>of</strong> course that a specification<br />

document can be understood and influenced by marketing experts. In general these<br />

people are not technology experts. The same holds for team members from various<br />

other expertise domains. They should be involved in the whole design path, starting<br />

from an adequate specification. In the existing specification tradition we observed a<br />

huge communication problem.<br />

Concurrent Approach<br />

There are few specification methods, for direct industrial application, that bridge the gap<br />

between hardware/s<strong>of</strong>tware design. A lot <strong>of</strong> research has been performed in the area<br />

<strong>of</strong> s<strong>of</strong>tware engineering and formal specification. A problem that we have with most<br />

<strong>of</strong> these approaches is that they propose to finish a specification phase before starting<br />

design. They assume that specification can be completed without performing design<br />

activities. The idea is that an optimal implementation technology will be chosen after the<br />

evaluation <strong>of</strong> alternatives. ’Implementation independence’ is elevated to an objective in<br />

itself. Apart from the fact that no one knows how orthogonal criteria must be weighed<br />

to create an economical optimum, this approach does not match with industrial practice<br />

in general. The very necessary practice is that some implementation decisions are taken<br />

partially beforehand or at least in an early phase. This is absolutely necessary to be<br />

able to reuse existing knowledge and technology, and because <strong>of</strong> the enormous time<br />

pressure that is imposed on product development nowadays. Concurrent engineering<br />

is necessary. There is no time for a sequential approach. Therefore we set ourselves the<br />

goal to integrate design partially with traditional implementation independent system<br />

modelling.<br />

System behaviour modelling is far from easy and time consuming. It is however necessary<br />

and pays <strong>of</strong>f at the end. Besides a lot <strong>of</strong> other reasons, modelling is simply necessary<br />

to find the required behaviour <strong>of</strong> a complex system to be designed. We experienced that<br />

a large part <strong>of</strong> the functionality <strong>of</strong> complex systems is only found when the behaviour<br />

is actually being modelled. It is just impossible to capture the complete functionality<br />

during a requirements phase, which is <strong>of</strong>ten recognised as preceding a modelling phase.<br />

For this reason promotion <strong>of</strong> the early use <strong>of</strong> specification models is necessary.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!