Enabling High-Quality Real-Time Communications with Adaptive Frame-Rate
summary
Adaptive Frame Rate (AFR) is a controller that coordinates frame rate with network fluctuation and decoder capacity to avoid transient decoder overloads, implemented for Tencent cloud gaming and validated with 40,000+ hours and 30,000+ sessions of traces. AFR addresses untimely queue signals and multiple queue causes, predicts delay from arrival/departure stats, and adapts frame rate (no codec change). Trace-driven sims and large deployments show 7.4× tail reduction and 34% fewer stalls and AFR is deployed in production as Adaptive Frame Rate (AFR).
AI-written summary · about
chapters (9)
- 0:00Introduction to RTC
- 1:00The problem of latency
- 2:52Decoder queue overload
- 5:06Adaptive frame rate solution
- 7:08Challenges in latency control
- 8:37Predictive queue modeling
- 10:53Addressing practical factors
- 11:44Evaluation and results
- 13:36Conclusion and takeaways
related talks
22mProviding better video experiences for the next billion users.
17mTooth: Toward Optimal Balance of Video QoE and Redundancy Cost by Fine-Grained FEC in...
16mFrom Source to Solution: Tackling Packet Losses in Large-scale Cloud Gaming...
03mImproving ABR at Netflix
19mUsing application informed pacing to be a friendly internet neighbor
15mMARC: Motion-Aware Rate Control for Mobile E-commerce Cloud Rendering
11mFree ABR Renditions for User Generated Content Platforms
13mUltra-High Resolution Playback Visualization and its Applications