ScholarWorks@숙명여자대학교

초록

This article proposes a novel two-stage framework for an advanced metaverse streaming with the objective of achieving high-quality and low-delay multiple virtual metaspace rendering. In the first stage, point cloud data obtained from physical space for rendering metaspace regions is considered, in which the scheduling for multiple metaspace regions is conducted. Among various scheduling algorithms, a novel quantum multiagent reinforcement learning (QMARL)-based scheduler is proposed due to its inherent ability to reduce the scheduling dimensions, making it highly beneficial for large-scale multiregion metaspace rendering. In the second stage, the gathered data must be rendered for visualizing multiple metaspaces in the metaverse server with the minimum number of processors for efficient operations. Then, the rendered multiple metaspace data are distributed to individual users with the objective of achieving high-quality multiple metaspace distribution under delay constraints. To support this objective, a server-side resource-optimal and low-delay scheduler is proposed.

키워드