论文部分内容阅读
分布式被动测量是研究网络行为的一个重要手段 .为了获得分布式协同处理流量信息 ,要求分布的测量点能抽取同样报文 ;为了能估计流量总体统计属性 ,抽样样本需要具有统计随机性 .为此 ,文章提出分布式掩码抽样测量模型处理高速网络流量 ,其核心是确定合适的抽样掩码匹配位串 .对CERNET主干网络流量IP报头各字段的位熵和位流熵进行分析 ,结果表明标识字段 16比特适合于抽样掩码匹配字段 .使用测量数据分析基于标识字段抽样模型的随机性和抽样样本的统计属性 ,实验进一步验证了所提出的模型具有良好的抽样性
Distributed passive measurement is an important means to study the network behavior.In order to obtain distributed collaborative processing of traffic information, distributed measurement points can extract the same message; in order to estimate the overall statistical properties of traffic, the sampling samples need to have statistical randomness. In this paper, we propose a distributed mask sampling measurement model to deal with high-speed network traffic, and its core is to determine the appropriate bit string for the match sample bitstream.Then we analyze the bit entropy and bitstream entropy of each field of CERNET backbone IP header, The 16 bits of the identification field are suitable for the sample mask matching field.Using the measurement data to analyze the randomness of the sampling model based on the identification field and the statistical properties of the sampling samples, the experiment further verifies that the proposed model has good sampling