摘要
针对网络流量的自相似、重尾分布等特征,对传统的系统抽样进行改进,设计出一种新的抽样方法——自适应系统双抽样。该算法以传统的系统抽样为基础进行改进,充分考虑了网络流量重尾分布的特点,能正确估算Hurst参数,实现简单,参数自适应且能控制资源消耗。通过真实网络数据的实验分析表明,在链路负载估计、包到达时间间隔等方面较传统抽样方法都有明显的改进,提高了测量系统的精确性和实用性。
A new sampling method-self adaptive systematic double sampling method is proposed, based on the self-similar and heavytailed distribution characteristics of the network flow. It＇s a revised version of traditional systematic sampling and can achieve simplicity, adaptability and controllability of resource consumption. This algorithm can fully consider the heavy-tailed distribution characteristic and estimate Hurst parameter correctly. The experiment analysis through the real network data indicates： Compared with traditional sampling methods, the new method has obvious improvement in link load estimate and packet interarrival times estimate, which advances the accuracy and practicability of the sampling measuring system.
出处
《计算机工程与设计》
CSCD
北大核心
2007年第22期5409-5410,5436,共3页
Computer Engineering and Design
基金
国防基础研究基金项目(A1420061266)
关键词
自相似
重尾分布
双抽样
链路负载
包到达时间间隔
self-similarity
heavy-tailed distribution
double sampling
link load
packet interarrival times
