基于混合遗传算法的输入向量控制漏电优化

Input Vector Control for Leakage Power Minimization Based on Hybrid Genetic Algorithm

提出了一种用于输入向量控制漏电优化的混合遗传算法。该算法首先根据输入信号对电路泄漏功耗的不同影响,将其划分为关键输入信号和非关键输入信号;接着将输入向量空间划分为子空间,混合遗传算法对子空间实施多样性控制实现全局搜索,在子空间内爬坡实现快速局部搜索。对采用HJTC(和舰科技)0.18μm工艺的ISCAS85电路的泄漏功耗进行了优化。结果表明,混合遗传算法的优化范围最大可达111.85%,而且其收敛速度更快。

A novel hybrid genetic algorithm （HGA）for input vector control on CMOS circuit was proposed for leakage power minimization. Taking advantages of primary inputs＇ different contribution to leakage power, the input vector space was divided into subspaces. With diversity control for critical primary inputs and bit climbing for minor primary inputs, the leakage power optimization based on HGA could be up to 111.85 % for ISCAS85 benchmark circuit in HJTC 0. 18 μm process.