Human-Centric Resource Allocation in the Metaverse Over Wireless Communications

Jun Zhao, Liangxin Qian, Wenhan Yu

TL;DR

This paper introduces a human-centric approach to resource allocation in the Metaverse, prioritizing user perception over traditional network metrics. By developing a novel mathematical technique for fractional programming, the authors successfully maximize the "Utility-Cost Ratio," balancing the subjective quality of VR video against the energy and latency costs of wireless delivery.

What Problem Does This Paper Address?

The Metaverse promises immersive experiences through Extended Reality (XR) and high-fidelity video streaming. However, delivering these experiences over wireless networks is a massive challenge. Traditional network optimization focuses on Quality of Service (QoS)—objective metrics like bit rate or packet loss. But in the Metaverse, the real judge is the human user.

The authors argue for a shift toward Quality of Experience (QoE). The challenge is that providing high-resolution, high-bitrate video to everyone consumes enormous amounts of energy and creates significant delays (latency), which can cause motion sickness in VR. This paper tackles the "balancing act" of deciding how much bandwidth, transmission power, and computing resource to give each user, alongside choosing the right video resolution, to maximize human satisfaction while minimizing system "pain" (cost).

Methods

The core of the paper is the formulation of the Utility-Cost Ratio (UCR) optimization problem.

  1. Human-Centric Utility: The authors modeled how humans perceive VR video quality as a function of data rate and resolution. They used logarithmic functions to represent the "diminishing returns" of quality—eventually, your eyes can't tell the difference, but the network certainly feels the load.
  2. System Cost: This includes the energy used by both the Metaverse Server (MS) and the VR Users (VUs), plus the total delay (processing time + transmission time).
  3. Novel Fractional Programming (FP): Mathematically, the UCR problem is "non-convex" and "sum-of-ratios," making it notoriously difficult to solve. The authors contributed a new FP technique that can handle objective functions containing a mix of convex functions and non-convex ratios.
  4. Algorithm A1: They developed a three-level iterative algorithm:
  5. Outer level: Uses Dinkelbach’s transform to handle the main ratio.
  6. Middle level: Uses Alternating Optimization to toggle between optimizing video resolution and other resources.
  7. Inner level: Employs their new FP technique to find a stationary point for the remaining variables.

Datasets

The authors didn't just guess how humans feel about video; they used real-world data to "learn" the utility functions:

  • SSV360 Dataset: Contains subjective quality assessments from users watching 360° videos on HTC Vive Pro headsets in different physical postures (standing vs. seated).
  • Netflix Dataset: Part of the VMAF project, providing mean opinion scores for various video bitrates and resolutions.

Key Insights and Ideas

  • The "Sweet Spot" of Resolution: The research found that the Utility-Cost Ratio often plateaus or even drops if resolution is pushed too high (e.g., moving from 1080p to 4K). In a resource-constrained wireless environment, "good enough" resolution for the human eye is often the most efficient choice for the system.
  • Separability of Resources: By using their FP technique, the authors showed that computation and communication resources can be optimized somewhat independently once certain "coupling" variables are fixed, simplifying a very complex problem.
  • Context Matters: The data showed that users' physical state (seated vs. standing) changes their perception of quality. Seated users tended to be more "forgiving" or comfortable, leading to higher UCRs in those scenarios.

Key Results

  • Superior Efficiency: The proposed algorithm significantly outperformed baselines. In simulations, it achieved a UCR improvement of 522% to 630% compared to a standard "average allocation" approach.
  • Latency Guarantees: The algorithm successfully kept per-frame delay under 7ms. This is crucial because VR generally requires a frame rate of at least 90 FPS (approx. 11ms per frame) to prevent user discomfort and motion sickness.
  • Robustness: The algorithm proved effective across varying numbers of users (from 10 to 160), though the average UCR naturally decreases as more users compete for the same server resources.

Limitations and Future Work

  • Stationary Points vs. Global Optima: Because the problem is non-convex, the algorithm finds a "stationary point" (a very good local solution) rather than a guaranteed global optimum. While the results are excellent, a global solution remains a theoretical "holy grail."
  • User Priorities: The current model treats all users' sum-utility equally. Future work could incorporate weighted priorities (e.g., a "premium" user getting better resources).
  • Real-world Implementation: The study is based on extensive simulations and real-world datasets, but the authors note that the next step is implementing this in a live, end-to-end wireless Metaverse testbed to evaluate real-time performance.

Summary updated

Abstract

The Metaverse will provide numerous immersive applications for human users, by consolidating technologies like extended reality (XR), video streaming, and cellular networks. Optimizing wireless communications to enable the human-centric Metaverse is important to satisfy the demands of mobile users. In this paper, we formulate the optimization of the system utility-cost ratio (UCR) for the Metaverse over wireless networks. Our human-centric utility measure for virtual reality (VR) applications of the Metaverse represents users’ perceptual assessment of the VR video quality as a function of the data rate and the video resolution and is learned from real datasets. The variables jointly optimized in our problem include the allocation of both communication and computation resources as well as VR video resolutions. The system cost in our problem comprises the energy consumption and delay and is non-convex with respect to the optimization variables. To solve the non-convex optimization, we develop a novel fractional programming technique, which contributes to optimization theory and has broad applicability beyond our paper. Our proposed algorithm for the system UCR optimization is computationally efficient and finds a stationary point to the constrained optimization. Through extensive simulations, our algorithm is demonstrated to outperform other approaches.
BibTeX
@article{zhao2024human,
  author = {Jun Zhao and Liangxin Qian and Wenhan Yu},
  title = {{Human-Centric Resource Allocation in the Metaverse Over Wireless Communications}},
  journal = {IEEE J. Sel. Areas Commun.},
  year = {2024},
  url = {https://openreview.net/forum?id=ysLZPVRt3i},
}

Loading related content...

Figure