QoS容忍下QoE的Logistic回归评估模型

用户感知质量(QoE)评估为有序多分类问题, 可以将Logistic回归模型应用于QoE的评估. 基于Logistic回归模型的基本原理, 建立QoE等级概率分布回归模型, 该模型能够完成服务质量(QoS)参数到用户QoE等级累积概率分布的映射. 提出考虑QoS容忍值的QoE评估方法, 该方法可以消除较好QoS指标对较差QoS指标的补偿作用, 在给出QoE所属等级的同时能够给出详细的QoE等级概率分布情况. 应用该方法对Web业务QoE进行了评估, 所提供的评估信息更详细, 评估结果更准确.

QoE Evaluation Method through Logistic Regression Model Considering QoS Tolerance

As the evaluation of quality of experience (QoE) is an ordinal multiple categories problem, the logistic regression model could be used to evaluate QoE. The regression model of QoE categories probability distribution is established based on the logistic regression model. This model relates cumulative probabilities of particular QoE categories to quality of service (QoS) parameters of network. Considering the tolerance of QoS parameters, a QoE evaluation method through QoE categories probability distribution regression model is proposed. This method eliminates the compensations among QoS parameters. It yields the category of QoE while giving a probability of each possible category. Finally, the QoE of Web service is estimated utilizing the proposed method and the result proves the method to be comprehensive and exact.