基于混沌优化算法的软硬件划分

Hardware/software partitioning by using chaotic optimization algorithm

针对软硬件协同设计中的关键问题———软硬件划分,提出一种基于混沌优化的划分算法.首先,使用有向无环图对嵌入式系统建模,得到软硬件划分优化系统的目标函数.然后,采用逻辑斯蒂映射产生混沌序列,并将此序列映射到划分系统的模型空间,利用混沌序列的遍历性,将粗搜索和细搜索相结合,分两阶段搜索模型空间目标函数的最优解,有效避免搜索过程陷入局部最小,并且使算法搜索时间大幅度降低.和模拟退火软硬件划分技术对比的实验结果表明,选取适当的算法参数,采用混沌优化算法能够以更快的搜索速度得到更好的软硬件划分结果.

Hardware/software partitioning is the key problem in hardware/software co-design. A new hardware/software partitioning algorithm of embedded system was introduced based on chaotic opti- mization algorithm. The model of the embedded system was constructed by directed acyclic graph to obtain the objective function for hardware/software partitioning. A chaotic sequence was generated by using Logistic mapping and mapped into modeling space of the partitioning system. With the chaotic sequence, both the coarse and fine searching were used to find the optimization of obiective function. The searching process was reduced largely without the local optimization and the searching time. The contrasting experiments to the simulated annealing algorithm show that, chaotic optimization algorithm with the reasonable parameters can accelerate the convergence and augment to find a good hardware/software partitioning.