CAJ | 학술논문

传统的包抽样方法对每一个数据包都以同等的比率抽取，这样就导致了大部分被抽中的是大流，而短流和一般流非常少。高速网络的流量检测需要全面的流信息。针对传统流抽样的缺陷，结合现有的SGS（ Sketch Guided Sampling）抽样比与流量成反比的公平抽样思想和动态计数型过滤器，提出更加高效的公平抽样算法DCFS（ Dynamic Count Fair Sampling）。 DCFS算法使用动态统计过滤器DCF（ Dynamic Count Filter）统计流量，相对于SGS算法该方法空间更加高效，而且估计准确性也更好。
전통적포추양방법대매일개수거포도이동등적비솔추취，저양취도치료대부분피추중적시대류，이단류화일반류비상소。고속망락적류량검측수요전면적류신식。침대전통류추양적결함，결합현유적SGS（ Sketch Guided Sampling）추양비여류량성반비적공평추양사상화동태계수형과려기，제출경가고효적공평추양산법DCFS（ Dynamic Count Fair Sampling）。 DCFS산법사용동태통계과려기DCF（ Dynamic Count Filter）통계류량，상대우SGS산법해방법공간경가고효，이차고계준학성야경호。
Traditional packet sampling method draws every packet at the same rate, which results in the vast majority of samples to be drawn are the large flows, but there are few the small and medium flows.Detecting the traffic in high-speed networks needs comprehensive flow information, for the defect of traditional sampling, in combination with the fair sampling idea of existing SGS ( sketch guided sampling) in which the sampling rate is in inverse ratio to traffic and with the dynamic count filter, we propose a more efficient fair sampling algorithm called DCFS ( dynamic count fair sampling ) .The DCFS algorithm applies DCF ( dynamic count filter ) to count the traffic, and is more efficient in space than the SGS algorithm, its estimation accuracy is also higher.