一种基于FCBF的流信息抽样测量框架及算法  

Architecture and Algorithm for Flow Information Sampling Measurement Based on FCBF

作　　者：张峰[1] 谭兴晔[1] 雷振明[1] 

机构地区：[1]北京邮电大学ATM技术研究中心,北京100876

出　　处：《计算机应用研究》2005年第6期38-41,共4页Application Research of Computers

基　　金：国家重大自然科学基金资助项目(69896240)

摘　　要：基于FCBF的高效流信息抽样测量框架不仅可以抽样测量三类流参数,而且存储开销小,只需1～3MB字节左右的存储空间;同时还可以做到几乎零概率的流信息识别统计误差。分析结果表明,该算法可以支持远高于OC48的链路速率,甚至可达OC192或更高;适合于将来高速链路上细粒度的流信息抽样测量。The architecture and algorithm which can sampling measure three types of flow information are proved that they are feasible to be implemented in hardware with low memory space required and little computation overhead, and can acquire the effect of almost zero probability on identifying and counting flow information working on a link faster than OC-48, even reaching OC-192 or even higher. So the architecture and algorithm adaptive to be implemented in current high-speed network link interface are suitable to measure flow distribution information with high resolution and fine granularity.

关 键 词：BF FCBF 流信息 抽样测量 开销 

分 类 号：TP393.07[自动化与计算机技术—计算机应用技术]