Decento: A New Scalable Interactive Live Streaming System via Control Plane Decentralization
Jongyun Lee, Sanghyun Han, Jeongmin Bae, Sangtae Ha, Kyunghan Lee
摘要
Large-scale interactive live video streaming enables real-time interaction among many streamers and viewers via video and message chat, requiring dynamic adaptation to viewers’ multi-view layouts and bandwidth conditions. Existing systems rely on SFU (Selective Forwarding Unit) servers and low-latency protocols such as WebRTC, but suffer from scalability limits: centrally aggregating all viewer states to determine bitrates incurs high computation and signaling overhead, leading to control delays that degrade responsiveness at scale. To address these challenges, we present Decento, a scalable interactive streaming system based on decentralized bitrate orchestration. Instead of centralizing viewer states, Decento lets each viewer submit preferred bitrate combinations to a nearby local control node, which aggregates them and selects representative combinations using an ϵ-covering algorithm to approximate global preferences. By forwarding only compact summaries to the central unit, Decento significantly reduces global reconfiguration complexity, suppresses signaling storms, and enables faster adaptation to viewer dynamics. Integrated with SFU servers, Decento achieves sub-3-second bitrate reconfiguration time at a 10,000-user scale — over 10× faster than state-of-the-art centralized systems — while improving average QoE by 29%.
问问这篇 Paper
问问你的智能体。
Lune 读过与它相关的顶会 Paper,每个回答都会注明依据哪几篇。
相关 Paper
- Exstream: A Delay-minimized Streaming System with Explicit Frame Queueing Delay MeasurementShinik Park, Sanghyun Han, Junseon Kim, Jongyun Lee 等INFOCOM 2024 · 被引用 1 次
- Scalable Video Conferencing Using SDN PrinciplesOliver Michel, Satadal Sengupta, Hyojoon Kim, Ravi Netravali 等SIGCOMM 2025 · 被引用 8 次
- MultiLive: Adaptive Bitrate Control for Low-delay Multi-party Interactive Live StreamingZiyi Wang, Yong Cui, Xiaoyu Hu, Xin Wang 等INFOCOM 2020 · 被引用 19 次
- An Intelligent Learning Approach to Achieve Near-Second Low-Latency Live Video Streaming under Highly Fluctuating NetworksGuanghui Zhang, Ke Liu, Mengbai Xiao, Bingshu Wang 等ACM MM 2023 · 被引用 5 次
- Harnessing WebRTC for Large-Scale Live StreamingWei Zhang, Tong Meng, Xianhua Zeng, Wei Yang 等SIGCOMM 2025 · 被引用 4 次
