摘要
流数估计是网络管控的重要参考尺度,对网络流量全局特征信息的深入挖掘具有重要意义。针对目前已有的多种估计算法以过度消耗测量设备存储资源和计算资源提高估计精度的缺陷,采用报文抽样技术,提出一种新的迭代收敛型估计算法。实验测试表明,该算法在估计精度和内存消耗上优于EM算法,在迭代更新上优于Iteration算法。
The estimation of flows number is an useful metric for network management and control, and has great significance _ w to the deeply mining traffic information in a network. Focusing on existing problems, such as excessive consumption of memory and computing resources for measurement, based on packet sampling, this paper introduced a novel convergent and iteration estimation algorithm. The experiment results demonstrate that the algorithm is superior to the expectation maximum(EM) algorithm on estimated accuracy and memory consumption, and superior to the Iteration algorithm on iterative updation.
出处
《计算机应用研究》
CSCD
北大核心
2015年第7期2078-2082,共5页
Application Research of Computers
基金
国家"973"计划资助项目(2012CB315901)
国家"863"计划资助项目(2011AA01A103)
关键词
流数
报文抽样
网络测量
收敛性
number of flows
packet sampling
network measurement
convergence