基于DTPSO算法的混合极性XNOR/OR电路功耗优化

摘要

通过对混合极性XNOR/OR展开式和CMOS电路动态功耗模型的研究,提出一种适合中大规模混合极性XNOR/OR电路的功耗优化算法.首先根据电路的静态逻辑实现形式,建立混合极性XNOR/OR电路功耗估计模型；然后改进快速列表技术以实现XNOR/OR电路的混合极性转换；再结合离散二值粒子群算法和机率转换法则,将离散三值粒子群优化(Discrete Ternary Particle Swarm Optimization,DTPSO)算法应用于混合极性XNOR/OR电路的最佳极性搜索,实现电路低功耗优化;最后对14个PLA格式MCNC Benchmark电路进行测试,结果表明:与已有的方法相比,该优化算法搜索到的混合极性XNOR/OR电路最佳极性,其对应的电路功耗平均节省53.98％,搜索速度得到明显提高.%A low power optimization algorithm for mid-large scale mixed polarity XNOR/OR circuits is proposed,by investigating the mixed polarity XNOR/OR expression and dynamic power consumptions of CMOS circuits.A mathematic mode for low power dissipation is established for mixed polarity XNOR/OR circuits according to the static logic form of circuit.A fast tabular technique is improved and applied in the polarity conversion of mixed polarity XNOR/OR circuits.Then combining discrete PSO and probabilistic transition rules,DTPSO is introduced to search the best polarity for power of mixed polarity XNOR/OR circuits.Finally,the proposed algorithm is tested on fourteen circuits from MCNC benchmark circuits,and the experimental results show that the algorithm is quite efficient in CPU time and have achieved average power-saving 53.98％,compared with the reported method.