M5: Facilitating Multi-user Volumetric Content Delivery with Multi-lobe Multicast over mmWave

Multi-user volumetric content delivery can enable numerous appealing applications, such as online education, telehealth, multi-user AR/VR training, immersive collaborative analytics, etc. However, the bandwidth-intensive nature of volumetric video streaming makes existing systems for single-user experiences hard to scale to multi-user scenarios. To address this critical issue, in this paper, we first perform a scaling experiment on mmWave networks that offer the needed multi-Gbps throughput and identify two key challenges of streaming high-quality volumetric videos to multiple users: frequent blockages of mmWave links and high transmission redundancy among users. To solve these problems, we propose a first-of-its-kind, agile, and cross-layer system, dubbed M5, for improving the performance and quality of experience for multi-user volumetric video streaming. M5 utilizes the 6DoF motion prediction of users to proactively adapt mmWave beams and prefetch frames to mitigate the blockage effects. Furthermore, it takes advantage of the multicast transmission to deliver the overlapped common content within users' viewports to reduce the bandwidth requirement. Our extensive experiments on a real testbed and with a trace-driven simulator show that M5 can effectively improve the frame rate by 44.1% and volumetric video quality by 62.3% compared to the state-of-the-art system.