Tooth: Toward Optimal Balance of Video QoE and Redundancy Cost by Fine-Grained FEC in...
summary
Paper analyzes over 60,000 cloud gaming sessions and finds coarse-grained FEC both over-protects large frames (wasting bandwidth) and under-protects small frames (causing stalls). It builds Tooth, a per-frame adaptive FEC with a dual-module FEC encoding (a fast lightweight ML mapper plus a long-timescale pattern updater) and deploys it in an operational system. Field tests show stall rates fall 40.2–85.2%, video bitrate rises 11.4–29.2%, and bandwidth cost drops 54.9–75.0%.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 1:01Latency requirements
- 2:47Packet loss handling
- 4:13Overprotection in FEC
- 6:19Confidence intervals for redundancy
- 7:44Variable frame sizes impact
- 9:29Frame-level loss distribution
- 11:35Fine-grained FEC per frame
- 12:58Modeling loss patterns
- 14:04Implementation and evaluation
- 15:10Performance comparison
- 16:14Conclusion
related talks
22mProviding better video experiences for the next billion users.
15mEnabling High-Quality Real-Time Communications with Adaptive Frame-Rate
16mSwift: Adaptive Video Streaming with Layered Neural Codecs
16mGRACE: Loss-Resilient Real-Time Video through Neural Codecs
33mCoding Theory for Scalable Media Delivery
16mFrom Source to Solution: Tackling Packet Losses in Large-scale Cloud Gaming...
16mMAE: More Adaptive Video Encoder for Consistent Low Latency in High-Quality Real-Time...
10mEnhancing CDN Performance and Cutting Egress Costs in Large Video Libraries
