Enhancing a Real-Time Distributed Computing Component Model through Cross-Fertilization

The need for overall optimization of software-hardware complexes has been there throughout the history of computer applications. However, the need for significant improvement in the techniques for achieving it has become very acute as the growth of the embedded computing application field has been in an accelerating mode since mid-1990's. As a natural consequence of it, the desire to have unified modeling approaches that are effectively applicable to both hardware systems and software systems is as strong as ever. Often software design activities and hardware design activities proceed largely independently without joint analysis and optimization, i.e., without disturbing each other, until the integration phase is reached. In such situations, software designers can benefit considerably from the availability of reliable models of hardware under development at the early stage. Similarly, hardware designers can benefit from the availability of early abstract but reliable models of software under development. Both designers are bound to wonder about the essential differences between hardware system modeling techniques and software system modeling techniques and about the aspects common to both types of modeling techniques. This author, who is a researcher dealing with techniques for software system modeling, feels that the following complimentary relationship exists between the recent work on software system modeling and that on hardware system modeling. • Software designs used to involve producing multiple layers of abstraction. In general, the number of layers defined during software