TABLE OF CONTENTS Embedded systems are usually evaluated according to a large variety of criteria such as performance, cost, flexibility, power and energy consumption, size and weight. As these kinds of non-functional objectives are very often conflicting, there is no single optimal design but a variety of choices that represent different design tradeoffs. As a result, a designer is not only interested in one implementation choice but in a well-chosen set that best explores these trade-offs.
In addition, embedded systems are often complex in that they consist of heterogeneous subcomponents such as dedicated processing units, application-specific instruction set processors, general-purpose computing units, memory structures and communication means like buses or networks. Therefore, the designer is faced with a huge design space.
Embedded systems are resource constrained because of tight cost bounds. Therefore, there is resource sharing on almost all levels of abstraction and resource types that makes it difficult for a designer to assess the quality of a design and the final effect of design choices. This combination of a huge design space on the one hand and the complexity in interactions on the other hand makes automatic or semi-automatic (interactive) methods for exploring different designs important.
Besides the above-mentioned multiple objectives in the design of embedded systems, there are tight constraints on the design time. One possibility to accommodate late design changes and a short time-to-market is to choose a very flexible design, close to a general-purpose computing system. On the other hand, this...
