Dissection of XR Physics Engine into a Cloud-native Service

Abstract

This paper addresses the demand for realistic physics simulations in collaborative Extended Reality (XR) experiences, particularly in gamified applications for untethered Head-Mounted Displays (HMDs). We propose optimizing the physics computations workload of modern game engines, by offloading computations to edge/cloud resources, serving multiple XR users in a 1-N fashion. Performance experiments with 25 users show improved rendering times under optimal network conditions, establishing thresholds for network latency and packet loss to maintain Quality of Experience (QoE) in XR applications. This approach enhances HMD battery life and user mobility, contributing to the advancement of realism and interactivity in XR environments.

Publication
Preprint