摘要
针对系统抽样中恒定报文采样概率方法异常检测的漏检率高和随机报文采样概率偏向于采集长流的缺陷,提出了一种模糊自适应流量抽样方法。该方法利用网络流量的相关性设定采样率,并实时自适应预测采样粒度,自动在允许误差范围内进行尽可能精确地测量,更好地捕捉到网络流量特征和网络行为特征。实验证明,所提方法不但能减少抽样数据对于异常检测的影响,而且可以高效地反映原始数据的异常情况。自适应模糊控制系统结构简单,易于实现,其控制策略达到了较高的工艺水平的要求,具有良好的预测性,并能提高流量检测的精确度,具有一定的推广价值。
According to correlation forecasting network traffic status,adaptive traffic sampling adjusts the sampling probability and granularity in real time,automatically achieves accurate measurement within the permissible error,to capture the network traffic characteristics and network behavior better.As undetected anomaly detection rate of constant probability sampling for packets is high,and the probability of random traffic sampling is apt to long stream,discuss a fuzzy adaptive traffic-sampling method.Experiments show the method can reduce the effect of sampling data on anomaly detection,and effectively reflect anomalies of the raw data.Adaptive fuzzy control system is simple and easy to implement.The control strategy has been achieved a high level of technology requirements,and good predictability.According to the social value,the control system can improve the flow of detection accuracy.
出处
《计算机技术与发展》
2012年第3期110-112,共3页
Computer Technology and Development
基金
国家自然科学基金资助项目(61072067)
高等学校学科创新引智计划(B08038)
关键词
流量预测
模糊控制
自适应抽样
traffic prediction
fuzzy control
adaptive sampling
