Wireless Multi-User Interactive Virtual Reality in Metaverse with Edge-Device Collaborative Computing

TL;DR

This work addresses the challenge of delivering wireless multi-user interactive VR within ultra-low latency constraints by introducing an edge-device collaborative computing framework that separates foreground and background rendering and leverages predictive background tiles. A Constrained Markov Decision Process is formulated and solved with the safe RL algorithm AQM-CUP, which incorporates Active Queue Management to prevent congestion and enforce the MTP latency threshold while balancing AoI and device power consumption. The approach includes a detailed system model, mmWave transmission assumptions, and a queue-based environment that supports joint decisions on rendering location, tile selection, and MEC resource allocation. Across simulations on real VR datasets, AQM-CUP demonstrates faster convergence, better QoE, reduced sensor information age, and lower device power consumption compared to baselines, highlighting practical benefits for immersive metaverse experiences. The work provides a concrete framework and algorithmic toolkit for deploying scalable, low-latency VR in future 6G MEC networks.

Abstract

The immersive nature of the metaverse presents significant challenges for wireless multi-user interactive virtual reality (VR), such as ultra-low latency, high throughput and intensive computing, which place substantial demands on the wireless bandwidth and rendering resources of mobile edge computing (MEC). In this paper, we propose a wireless multi-user interactive VR with edge-device collaborative computing framework to overcome the motion-to-photon (MTP) threshold bottleneck. Specifically, we model the serial-parallel task execution in queues within a foreground and background separation architecture. The rendering indices of background tiles within the prediction window are determined, and both the foreground and selected background tiles are loaded into respective processing queues based on the rendering locations. To minimize the age of sensor information and the power consumption of mobile devices, we optimize rendering decisions and MEC resource allocation subject to the MTP constraint. To address this optimization problem, we design a safe reinforcement learning (RL) algorithm, active queue management-constrained updated projection (AQM-CUP). AQM-CUP constructs an environment suitable for queues, incorporating expired tiles actively discarded in processing buffers into its state and reward system. Experimental results demonstrate that the proposed framework significantly enhances user immersion while reducing device power consumption, and the superiority of the proposed AQM-CUP algorithm over conventional methods in terms of the training convergence and performance metrics.

Figures (11)

• Figure 1: Grid locations in the virtual world. Spatiotemporal VR panorama at location point is divided into FoVs. Merge background tile in the base layer and foreground objects in the enhancement layer to generate viewport frame.
• Figure 2: An illustration of the system model at the $k$-th time slot. The left side depicts CG rendering steps. The right side illustrates wireless multi-user VR with edge-device collaborative computing.
• Figure 3: The workflow of interactive VR with edge-device collaborative computing.
• Figure 4: Timing sequence of control decisions and the corresponding rendering, compressing, transmitting, decompressing and merging tasks in the interactive VR with edge-device collaboration. For simplicity of presentation, (a) all processing queues are empty before the $k$-th time slot, (b) the obtainable window length for the background $L=5$, (c) $\tau$ is set to half of $T^{\text{MTP}}$, i.e., $T^{\text{MTP}} = 20$ ms and $\tau = 10$ ms, corresponding to 100 FPS.
• Figure 5: Illustration of proposed AQM-CUP architecture.