摘要
针对系统约束下的片上网络映射如何建立低功耗和链路负载的多目标优化函数,提出一种基于融合离散粒子群算法(Discrete Particle Swarm Optimization Algorithm,DPSOA)和遗传算法(Genetic Algorithm,GA)的新型映射算法.该算法利用任务节点通信量大小及其连接关系,划分优先级,得到若干较优初始解集;利用离散粒子群算法的快速搜索能力迅速靠近最优解,利用遗传操作中的选择和变异防止算法掉入局部较优解陷阱,以较少的迭代次数完成最优解的寻找.实验结果表明:与遗传算法、粒子群算法和蚁群算法相比,该算法在功耗和链路负载优化上都能达到较好的结果.
In order to build a multi-objective function of low power and link load in NoC （ Network-on-Chip ） mapping with the consideration of the systematic limitations,a novel heuristic comb＇mation algorithm,based on Discrete Particle Swarm Optimization Algorithm （DPSOA） and Genetic Algorithm（GA） ,was proposed in this paper. This algorithm prioritized the nodes according to their communication weights, and then acquired a better initial mapping solution set through the task node priority and their connection. Finally, it obtains the optimal solution at the advantage of DPSOA on its rapid search process. Moreover, we introduced the choice and mutation of GA to prevent the algorithm stagnation,and finally got the global optimal solution within fewer iterations. Experimental results show that, compared with the GA, DPSOA and ACA （Ant Colony Algorithm ）, our proposed algorithm achieved better results on power consumption and link load.
出处
《小型微型计算机系统》
CSCD
北大核心
2017年第3期651-656,共6页
Journal of Chinese Computer Systems
基金
国家"八六三"高技术研究发展计划项目(2014AA01A)资助
关键词
片上网络
低功耗
链路负载
离散粒子群算法
遗传算法
Network-on-Chip
low-power
link load
discrete particle swarm optimization algorithm
genetic algorithm
