A Networking Perspective of Volumetric Video Service: Architecture, Opportunities, and Case Study

Yili Jin, Junhua Liu, Kaiyuan Hu, Fangxin Wang

Abstract: Recent years have witnessed the prompt development of multimedia applications over the Internet, where volumetric video stands out as a thrilling new video paradigm, leading a revolution from traditional flat 2D to the future volumetric 3D experience with spatialized immersion and 6 degree-of-freedom interactivity. It serves as a key technical foundation of AR, VR, MR, and the emerging Metaverse. In this paper, we mainly study the volumetric video service from a networking perspective. We start with a comprehensive overview of the volumetric video, clarifying its evolution and unique features. We then present a holistic view of the architecture of volumetric video service from a networking perspective, including video capture, representation, streaming, and display. Research opportunities and challenges are next discussed. Finally, we present our design on volumetric video streaming as a case study. The proposed framework achieves accurate 6 DoF viewport prediction while employing deep reinforcement learning for adaptive streaming. The evaluation demonstrates its effectiveness, exhibiting a nearly 30% improvement compared to existing systems.