基于先验概率的性能降级链路推断方法  

Inference Method for Lossy Links Based on Prior Probability

作　　者：赵佐[1] 蔡皖东[1] 

机构地区：[1]西北工业大学计算机学院,西安710129

出　　处：《计算机工程》2010年第3期13-15,18,共4页Computer Engineering

基　　金：教育部博士点基金资助项目(200806990030);西北工业大学科技创新基金资助项目(2008KJ02028)

摘　　要：研究基于布尔网络断层扫描技术辨识网络中性能出现严重降级的链路。将辨识问题归结为最大概率解释问题,提出一种利用链路状态先验概率分布的推断性能严重降级链路的方法。通过计算每种候选链路指派的后验概率,从中选取具有最大概率的指派作为推断结果。理论分析证明了该方法的可行性,仿真实验验证了该方法的有效性和准确性。This paper studies identification of lossiest links based on Boolean Network Tomography. It addresses this problem of identification of lossiest links to the most probable explanation in Bayes estimation. A method using link state prior probability is presented to inferred lossiest network links, the presented method computes posteriori conditional probability of each candidate link assignments and chooses the assignment with maximum probability. Theoretic analysis proves that the method is feasible, and simulation results show that the method achieves accuracy higher than previously proposed heuristic methods.

关 键 词：网络断层扫描 链路状态先验概率 最大概率链路指派 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]