基于遗传算法的并行网络模拟拓扑划分  

Topology Partitioning of Parallel Network Simulation Based on Genetic Algorithm

作　　者：王晓锋[1] 毛力[1] 

机构地区：[1]江南大学物联网工程学院轻工过程先进控制教育部重点实验室,江苏无锡214122

出　　处：《计算机工程》2011年第23期83-85,共3页Computer Engineering

基　　金：江南大学轻工过程先进控制教育部重点实验室开放课题基金资助项目(APCLI1004);中央高校基本科研业务费专项基金资助项目(JUSRP11131)

摘　　要：要提高并行网络模拟性能,需对网络模拟拓扑进行有效划分。为此,提出一种并行网络模拟拓扑的优化划分方法。分析影响并行网络模拟性能因素,给出并行网络模拟性能估计模型,以该模型为评价函数,采用遗传算法寻找优化划分,实现并行网络模拟拓扑的优化划分。在PDNS上的实验结果表明,与传统划分方法相比,该优化划分方法的并行模拟性能平均提高13.3%。To improve the performance of parallel network simulation,the topology of network simulation should be partitioned reasonably.On the basis of this,this paper proposes an optimized partitioning method of parallel network simulation topology.It analyzes key performance factors that affect the performance of parallel network simulation,gives a model for estimating the performance of parallel network simulation,the optimized partitioning of parallel network simulation is realized by treating this model as evaluating function,and the optimal partitioning by genetic algorithm is searched.Through experiments in PDNS,result shows that the performance of parallel simulation the optimized partitioning method improves by 13.3% in average compared with the traditional partitioning method.

关 键 词：网络模拟 并行网络模拟 模拟性能 性能估计模型 拓扑划分 遗传算法 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]