QoE estimation models for tele-immersive applications

Tele-immersive applications, which are regarded as the next generation distributed multimedia applications, are highly interactive and aims to offer an immersive experience to its users. A challenge with such applications is to provide the best possible quality of experience (QoE) under changing conditions. In particular, it would be highly desirable to be able to predict the QoE perceived by users in response to adaptation, prior to actual deployment. However, there are no QoE prediction models for tele-immersive applications. Instead QoE is evaluated after deployment using either objective or subjective assessment techniques. Unfortunately, objective assessment lacks the accuracy of human perception. At the same time, subjective assessment requires human-provided ratings of the applications, which is time consuming and thus not cost-effective. In this paper, we propose QoE prediction models that will accurately predict the user-perceived QoE of a tele-immersive conferencing application. The proposed models are cost-effective and lend themselves to fast evaluation cycles, because the models does not involve human-provided ratings. We validate our models using results from subjective assessment experiments. The models can be used for real-time monitoring of user-perceived QoE, in addition to designing QoE-driven adaptation for tele-immersive applications.