基于信息融合的多元QoS监控方法

摘要

QoS(Quality of Service)是衡量第三方服务质量的重要标准,有效监控QoS是Web服务质量控制的必要过程.针对现有的QoS监控方法几乎只考虑单个QoS指标,无法满足用户的满意度要求这一问题,提出了一种基于信息融合的多元QoS监控方法,对不同的QoS属性样本进行归一化处理,使用信息融合方法将多个QoS属性样本融合为综合QoS数据样本.再提取样本的环境因子,通过PMI (Pointwise Mutual Information)算法计算环境因子的分类倾向,构造加权朴素贝叶斯分类器,监控时调用分类器分析得到监控结论.在网络开源数据集以及模拟的数据集上的实验结果表明,与现有方法相比监控效果显著提高.%QoS (Quality of Service) is an important criterion for the quality of the third party service.Effective monitoring of QoS is the necessary process of Web service.Aiming at the problem that the existing QoS monitoring approaches only consider a single QoS index and can not meet the requirements of the user's satisfaction,a multivariate QoS monitoring approach based on information fusion is proposed.The approach first normalizes the different QoS attribute data,and uses the information fusion to integrate multiple QoS attribute samples into integrated QoS data samples.Then the weighting naive Bayesian classifier is constructed by using PMI (Pointwise Mutual Information) algorithm to calculate the classification tendency of feature factors.The monitoring conclusion is got by calling the classifier and analyzing the classification results.The experimental results on the network open source dataset and the simulated data set show that the approach significantly improves the effects compared with the existing approaches.