基于改进的NSGA遗传算法的SOC软硬件划分方法

摘要

An advanced non-dominated sorting genetic algorithm (ANSGA) is proposed, by introducing a removing method for building Pareto non-dominated sets (NDS) and an elitism preserving strategy for generating NDS and new sets,the ANSGA has a global convergence and reduced computational burden. The ANSGA was applied to hardware/software partitioning in the system on a chip (SOC) design. An integer vector coding solution was used for the individual coding, which is for the multi-intellectual property cores SOC design.Some experiment results are given,which show that the proposed method is superior and effective.%在遗传算法中引入精英保持策略构造非支配集和生成新群体,并用排除法构造非支配集,本文得到了一种改进的非支配集分类遗传算法,该算法具有全局收敛特性,并简化了计算复杂性.本文将此算法应用于SOC设计的软硬件划分.针对SOC系统设计中存在多个IP核的特点,采用整数向量的个体编码方案,避免了个体编码解码的冗余.本文给出了仿真实验结果,验证了该划分方法的有效性.