基于随机扫描的并行网络模拟拓扑划分算法  

A topology partition algorithm for parallel network simulation based on random scanning

作　　者：张兆心[1,3] 杜跃进[1,2] 王克[1] 童琳[1] 郝志宇[3] 

机构地区：[1]哈尔滨工业大学计算机科学与技术学院,哈尔滨150001 [2]国家计算机网络应急技术处理协调中心,北京100029 [3]中国科学院计算技术研究所,北京100190

出　　处：《高技术通讯》2012年第6期604-609,共6页Chinese High Technology Letters

基　　金：863计划（2007AA010503）,国家自然科学基金（61100189,61003261）和山东省中青年科学家奖励基金（BS2011DX001）资助项目.

摘　　要：为了提高并行网络模拟的性能，研究了实现有效的拓扑划分的策略，提出并实现了基于随机扫描的并行网络模拟拓扑划分（TPBRS）算法。基于启明星辰探测获得的实际拓扑进行的蠕虫模拟表明，该划分方法可适用于实际拓扑，并可进行大规模网络安全事件的模拟。实验结果表明，相对于传统划分算法，该拓扑划分方法减少模拟时间约19％，各个模拟节点模拟时间差值平均减少约21．78％，内存差值平均减少约4．6％，并且模拟时间和内存的增长更具有规律性，即负载均衡度更好，划分更加合理，提高了网络模拟的性能。The study aimed to find an effective topology partition policy to improve the performance of the parallel network simulation. With the theory of random scanning probability, an algorithm for topology partitioning based on random scanning （TPBRS） for parallel network simulation was put forward and implemented. The simulation results based on the real topology obtained by Venusense prove that this partition algorithm can be used in real topology and large scale security event simulations. The test results prove that compared with the traditional partition, this partition policy can reduce the simulation time by 19 percent, the difference of simulation time among simulation nodes by 21.78 percent, and the memory by 4.6 percent. The growth of simulation time and memory have good regularity, which means the better load balancing degree and partition result, and improvement of the performance of simulation.

关 键 词：并行网络模拟 拓扑划分 随机扫描 概率 

分 类 号：TP393.01[自动化与计算机技术—计算机应用技术]