QoE-Based Scene Adaptation Scheme of Distributed XR Applications

Distributed eXtended Reality (XR) applications are prone to network delay fluctuations that significantly degrade QoE. This issue can be addressed using scene-adaptation techniques, particularly those that dynamically adjust the Level of Detail (LoD). This work presents an LoD-based scene adaptation mechanism designed to enhance the QoE of XR applications under fluctuating network delays. An adaptation event occurs in the XR application whenever the observed QoE deviates from a predefined interval. The QoE level is derived from the PTRP (Pose-to-Render-to-Photon) metric collected at the XR client side (i.e., the user device). Throughout an adaptation event, the XR client and server coordinate gradual scene changes to achieve the desired QoE level. Specifically, the XR client requests either an increase or decrease in scene rendering time depending on the observed QoE level. On this basis, the XR server modifies the rendering complexity of the scene. This progressive and joint optimization process ensures optimal QoE despite rapidly changing network conditions. Experimental results indicate that the proposed scene adaptation scheme improves both instantaneous and overall QoE level under diverse network conditions. Moreover, we observe that our scheme enable to maintain a stable QoE level during XR sessions.