链路状态的概率分布作为先验知识对于推断链路性能状态的准确度起着重要作用.文中主要研究了在树形拓扑下基于端到端测量的内部链路状态概率推断问题,并将该问题定义为极大似然估计问题.采用乘积模型描述路径与链路的状态概率之间的关系,将链路状态概率的推断归结为路径状态概率的估计,提出了一种通过计算路径状态概率进而获得链路状态概率的方法,并将该方法用于仿真实验.结果表明,该方法具有较高的有效性和实用价值,能够准确有效地推断网络内部链路状态概率.

As a kind of prior knowledge,link state probability distribution plays an important role in inferring the accuracy of link performance state.This paper deals with the inference of internal link state probability based on the end-to-end measurement in tree topology,and defines it as a maximum likelihood estimation problem.By using a product model to describe the relationship between the path and the link state probability and by estimating link state probability via the computation of path state probability,a new approach to the fast inference of link state probability is proposed.The approach is then applied to simulation experiments.The results indicate that the proposed approach is effective and practical in inferring the internal link state probability.