主要研究了基于CR(capture-recapture)模型业务流估计算法CARE的过程和模型,分析了区分水平值、估计次数和数据列表长度对CARE业务流估计结果的影响,提出一种改进CARE业务流估计算法。新算法修改了CARE算法的估计过程,通过递归逐渐得到稳定的估计结果。实验分别采用计算机模拟数据和美国应用网络研究国家实验室NLANR的被动测量和分析工作组(PMA)的数据来对两种算法进行比较分析。结果表明改进的CARE算法业务流估计更准确。改进的CARE算法可以应用于网络中间设备实现业务流数目的估计,保证公平的带宽分配,具有很广泛的应用前景。
Through researching estimating-process and the model of CARE algorithm,which based on the model of CR. This paper analyzed effects of parameters capture-list length, estimating-times and difference-value, advanced an improved CARE algorithm for estimating traffic flow numbers. New algorithm changed the estimating-process of CARE algorithm and got the stable result by recurring gradually. The experiments compared and analyzed two algorithms through computer simulation dataset and passive Measurement and Analysis(PMA) of the National Laboratory for Applied Network Research (NLANR) dataset, which indicated that the improved CARE algorithm estimated flow number more precisely. The improved CARE algorithm can be applied in middle network devices to implement estimating flow numbers, ensuring bandwidths allocated fairly, which has extensive future of applications.