基于人工蜂群算法的低能耗高性能NoC映射  

Artificial bee colony based low-energy consumption high-performance mapping in NoC

作　　者：邓植[1] 顾华玺[1] 杨银堂[2] 曾代兵[3] 

机构地区：[1]西安电子科技大学综合业务网理论及关键技术国家重点实验室,陕西西安710071 [2]西安电子科技大学微电子学院,陕西西安710071 [3]中兴通讯股份有限公司,广东深圳518057

出　　处：《西安电子科技大学学报》2012年第2期114-119,共6页Journal of Xidian University

基　　金：国家自然科学基金资助项目(60803038,60725415,61070046);中央高校基本业务费资助项目(K50510010010);高等学校学科创新引智计划资助项目(B08038);国家重点实验室专项基金资助项目(ISN1104001);中兴产学研资助项目

摘　　要：针对网格型拓扑网络中心容易产生热点及高能耗问题,通过分析片上网络(NoC)结构,建立低能耗高性能多目标映射模型,提出了基于人工蜂群的IP核映射算法进行多目标求解.该算法通过模拟不同蜜蜂的操作以分别获取低能耗和流量均衡的映射解集,并采用交集选优策略完成最优映射的选取.仿真结果表明,相比现有算法,提出的映射算法不仅可以快速完成IP核到网络节点的映射,而且产生的映射结果在能耗、时延和流量均衡方面具有优势.In order to overcome the hotspot problem in the center of the mesh network and high energy consumption,an Artificial Bee Colony（ABC） based mapping algorithm is proposed to solve the low-energy consumption high-performance mapping mode,which is set up by the analysis of the Network-on-Chip（NoC） architecture.The algorithm can lead to a low energy consumption and balanced traffic optimal mapping set by simulating the operations of different bees.The result can be achieved by selecting the optimal mapping from the intersection of the solution set.Simulation results show that the proposed mapping not only accomplishes the mapping from the IP core to node at a fast rate,but also has better performance in energy consumption,latency and traffic balance.

关 键 词：片上网络 面向应用 映射优化 

分 类 号：TN47[电子电信—微电子学与固体电子学]