Abstract:
Most Quality of Experience (QoE) studies report only the mean opinion scores (MOS) and existing models typically map Quality of Service (QoS) parameters to the MOS. Howev...Show MoreMetadata
Abstract:
Most Quality of Experience (QoE) studies report only the mean opinion scores (MOS) and existing models typically map Quality of Service (QoS) parameters to the MOS. However, service providers may be interested in the share of users that are not at all satisfied, and their corresponding QoE levels. From the QoE management point of view, the circumstances leading to the QoE levels perceived by a certain percentage of users, e.g. the 10% most annoyed users, are of utmost importance. Proper metrics are the 10%-quantiles of QoE values. Knowledge of those quantiles helps service providers to estimate the need for countermeasures in order to prevent annoyed users from churning on one hand, and to avoid overprovisioning on the other hand. The contribution of this paper is the derivation of quantiles from existing MOS-QoS relations. This allows to reuse existing subjective MOS results and MOS models without rerunning the experiments. We consider exemplary the IQX model (describing the MOS-QoS relation) for the derivation of the quantile-QoS relation. A practical guideline for the computation of the quantiles is provided.
Date of Conference: 08-12 May 2017
Date Added to IEEE Xplore: 24 July 2017
ISBN Information: