Optimal component selection for energy-efficient systems

Abstract

Microelectronics have developed very fast in the past. The design process of those systems is getting more and more complex and new design methods have to be applied continuously. One main observation is the increasing design level over time, hence, the re-use of components is getting more important. A key challenge of IC design is the selection of a system architecture which fulfills all requirements in terms of data throughput, area, timing, power and cost. We present a problem class for the optimal component selection in order to assist in selecting the best available alternatives. We will show how to express top-level constraints and optimisation targets including dependencies between components. In addition, heuristic solving algorithms will be presented. The evaluation section shows that the presented algorithms perform well on typical problem sets. Using a framework for evolutionary algorithms results in additional speedup. © 2015 Springer International Publishing.