| 一种适用于流匹配的多阶段无冲突归并算法 |
| |
| 引用本文： | 吴勇军 唐红. 一种适用于流匹配的多阶段无冲突归并算法[J]. 重庆邮电学院学报(自然科学版), 2005, 17(3): 360-364 |
| |
| 作者姓名： | 吴勇军 唐红 |
| |
| 作者单位： | 重庆邮电学院 重庆400065(吴勇军)，重庆邮电学院 重庆400065(唐红) |
| |
| 基金项目： | 重庆市科技攻关项目(7220-13-20)，重庆市教委科技项目(020509)，重庆市科委自然科学基金资助 |
| |
| 摘 要： | 流匹配算法是准确实现对高速网络中特定流的流量进行测量的关键。RFC算法是相关算法中速度最快的，但它内存消耗很大，预处理运算量大。针对BFC存在的不足，提出了一种多阶段无冲突归并流匹配算法，该算法采用标记桶方法实现前缀匹配的预处理；采用最小区间分隔法实现范围匹配的预处理，同时算法设计了提前丢包。测试结果表明该算法能支持掩码匹配和范围匹配，内存消耗可控，匹配速度可达2Mpps。
|
| 关 键 词： | 流量测量 流匹配 多阶段无冲突归并 RFC |
Multi-phase non-collision merge algorithm for flow matching |
| |
| WU Yong-jun,TANG Hong. Multi-phase non-collision merge algorithm for flow matching[J]. Journal of Chongqing University of Posts and Telecommunications(Natural Sciences Edition), 2005, 17(3): 360-364 |
| |
| Authors: | WU Yong-jun TANG Hong |
| |
| Abstract: | Flow matching is one of the key issues of specificly based flows traffic measurement. RFC algorithm is the fast one among related algorithms, but the amount of memories needed and preprocessing computation is extremely huge. To overcome the disadvantage of RFC, the authors present an improved algorithm, namely multi-phase non-collision merge flow matching algorithm, which performs the prefix matching by marker-packet and range-matching by minimum range splitting. And it deploys the early discarding packet policy. The testing result shows the matching speed of this algorithm is up to 2Mpps and the amount of used memory is under control. The algorithm can also support mask matching and range matching. |
| |
| Keywords: | traffic measurement flow matching MPNCM RFC |
| 本文献已被 CNKI 维普 等数据库收录！ |
