Hierarchical optimization of multi-objective embedded system design using Pareto-frontier algebra

作者：Han Muhua^*; Liu Leibo; Wei Shaojun

来源：Chinese Journal of Electronics, 2008, 17(3): 402-408.

• embedded system
• multi-objective
• hierarchical optimization
• Pareto-frontier

This paper presents a novel scheme of hierarchical design space exploration, aiming at multi-objective embedded system design. Multi-objective hierarchical design methodology is prospective to deal with the design of modern complex embedded systems. However, the impact of decision data piling up across hierarchy boundaries as occurred in the previous works has made this method inefficient and restricted. This paper employs Pareto-optimal theory to analyze the hierarchical design space, and indicates how intermediate solutions of lower levels contribute to the final optimal solutions. By these indications, the paper adopts a multi-level treelike optimization procedure, and has proved that, under the independence conditions, optimization in each hierarchical level can be performed independently. Further more, the frameworks of unconstrained and constrained hierarchical optimization systems have been provided and proved. This is the very first time to formally explore the design space of multi-objective hierarchical system, which contributes to the promotion of novel hierarchical partitioning and synthesizing methodology.

• 出版日期2008-7
• 单位清华大学