Automatic model based design space exploration for embedded systems

Decades of electronic system design have shown design automation being the key to higher productivity. After automatic placement and routing for transistor layout, the next step was logic and high level synthesis converging to the system level where an algorithmic problem description has to be executed by resources that are unknown in advance. Since designing a complex system requires the consideration of many objectives and constraints simultaneously, one of the most challenging tasks at the system level is the so-called design space exploration. Design space exploration is the tasks of systematically exploring tradeoffs between different design goals and finding the optimal solutions. This thesis presents the state of the art in automatic model-based design space exploration as well as novel approaches in modeling, analysis, and optimization of complex embedded systems. The key contributions of this work can be stated as follows: A new design space representation is proposed allowing to model dynamic hardware reconfiguration as well as design reuse. New strategies in analyzing embedded systems on the basis of symbolic techniques are investigated. Finally, novel methods for accelerating design space exploration by using hierarchical decomposition and Multi-Objective Evolutionary Algorithms are presented.