Title :
HW-SW partitioning based on genetic algorithm
Author :
Zou, Yi ; Zhuang, Zhenquan ; Chen, Huanhuan
Author_Institution :
Dept. of Electron. Sci. & Tech, USTC, Anhui, China
Abstract :
HW-SW partitioning is an important problem in HW-SW codesign of embedded systems. We establish a HW-SW partitioning model based on system´s basic scheduling block (BSB) graph and propose a modified genetic partitioning algorithm (MGPA). By adopting an adaptive fitness function definition and a novel evolving strategy, we enhance the stability, efficiency and result quality of our partitioning algorithm. Experiment results show the algorithm´s effectiveness in solving the HW-SW partitioning problem.
Keywords :
computational complexity; embedded systems; genetic algorithms; graph theory; hardware-software codesign; scheduling; system-on-chip; HW-SW codesign; HW-SW partitioning; NP-hard problems; adaptive fitness function; basic scheduling block graph; control data flow graph; embedded systems; genetic algorithm; modified genetic partitioning algorithm; system-on-chip; task graph; Consumer electronics; Cost function; Design methodology; Embedded system; Energy consumption; Genetic algorithms; Hardware; Modems; Partitioning algorithms; Resource management;
Conference_Titel :
Evolutionary Computation, 2004. CEC2004. Congress on
Print_ISBN :
0-7803-8515-2
DOI :
10.1109/CEC.2004.1330916
