基于内部监测器的链路丢包率的推测方法  

Inference method for link loss rates with internal monitors

作　　者：苏海波[1] 林世俊[1] 陈文涛[1] 金德鹏[1] 曾烈光[1] 

机构地区：[1]清华大学电子工程系,微波与数字通信国家重点实验室,北京100084

出　　处：《清华大学学报（自然科学版）》2009年第8期1189-1192,1196,共5页Journal of Tsinghua University(Science and Technology)

基　　金：国家“九七三”重点基础研究发展计划资助项目(2007CB310701);国家“八六三”高技术项目(2008AA01A3312008AA01Z107)

摘　　要：为了能简单准确地估计网络内部链路的丢包率,该文利用网络层析技术提出了一种低复杂度的推测方法LCIA(low complexity inference approach)。LCIA方法通过在网络内部节点部署监测器来减小推测丢包率的复杂度。它主要由两步组成,第一步是在网络内部部署监测器把整棵树分成较小深度的若干棵子树。第二步是估计每棵子树的链路丢包率,用到的估计式是叶子节点观测结果的显式表达式来估计丢包率。定性的分析和实验仿真结果表明LCIA在计算复杂度和估计结果的精确度上,都优于已有的方法。A low complexity inference approach, LCIA, was developed to simply and accurately infer link loss rates. The system deploys monitors at internal nodes to reduce the complexity of inferring the link loss rates. The system first deploys monitors at specific internal nodes to divide the original tree into several sub-trees. Then the system infers the link loss rates in the sub-trees by an estimator, which is an explicit function of loss measurements. Analyses and simulations demonstrate that the method outperforms existing methods both on computational complexity and inference accuracy.

关 键 词：网络层析 链路丢包率 内部监测器 

分 类 号：TP393.0[自动化与计算机技术—计算机应用技术]