Quality of Service Impact on Edge Physics Simulations for VR.

Area of Science:

• Computer Science
• Virtual Reality
• Edge Computing

Background:

• Mobile Head-Mounted Displays (HMDs) face performance limitations due to power and ergonomic constraints.
• Current solutions like render streaming for HMDs are challenged by strict latency requirements for head tracking.
• Advancements in mobile GPUs necessitate exploring new methods to balance computational load.

Purpose of the Study:

• To propose and evaluate an alternative to render streaming for mobile VR called edge-physics.
• To investigate the feasibility of streaming at the scene-graph level from edge resources.
• To optimize the balance between computational load, latency, and visual quality in mobile VR.

Main Methods:

• Developed a prototypical implementation of the edge-physics system.
• Evaluated the system's performance based on fidelity, bandwidth consumption, and processing speed.
• Compared scene-graph streaming to traditional render streaming concepts for HMDs.

Main Results:

• Demonstrated that an effective edge-physics system can be built for typical edge network links.
• Showcased the compute and bandwidth efficiency of scene-graph level streaming.
• Identified that some traditional rendering concepts are not directly applicable to this approach.

Conclusions:

• Edge-physics offers a viable method for offloading demanding computational tasks from mobile VR devices.
• Maximizing compute utilization by offloading CPU-intensive jobs and leveraging mobile GPUs is achievable.
• Further research into the perception of motion is crucial for comprehensive evaluation of edge-physics systems.