通过分析Internet的本地聚集特性,给出了local-area和connect-area的定义,并基于此,为提高并行网络模拟性能,提出一种新型拓扑抽象算法——基于local -area的拓扑抽象(TABLA)算法.TABLA算法在给定的聚合粒度下,迭代搜索网络内的local-area,对拓扑进行抽象.模拟结果表明在Internet路由级拓扑上采用该算法,拓扑规模大约压缩为原先的45%,初始化内存节省约60%,模拟运行时间约缩短80%,大幅度提高了并行网络模拟性能.%This paper gives the definitions of local-area and connect-area based on the analysis of the local clustering property of the Internet, and on the basis of this, proposes the TABLA, a novel topology aggregation algorithm based on local area to improve the performance of parallel network simulation. The algorithm searches the local-area through the iteration method in a designated clustering level to perform topology aggregation. The simulation results show that for the Internet router-level topology aggregation, this algorithm can reduce the topology size by 45% , lessen the memory by 60% and shorten the simulation time by 80%.
展开▼