Mortise: Auto-tuning Congestion Control to Optimize QoE via Network-Aware Parameter
summary
Mortise auto-tunes rule‑based congestion control in real time by inferring an application's throughput‑delay‑loss preference and mapping that to CCA parameters, implemented on Linux TCP with eBPF + Copa delta tuning. It introduces a QoS tradeoff proxy that extracts lambda/beta from instantaneous QoE gradients, uses adaptive measurement and a feedback controller to converge (demo shows near‑optimal in <1s), and shows production gains across 128 cities with low overhead, reporting up to 73% and 167% QoE gains for downloads and streaming.
AI-written summary · about
chapters (8)
- 0:00Mortise framework
- 1:43Congestion control for QoE
- 4:12CCA trade-offs
- 6:20Current solution limits
- 8:05Mortise challenges
- 11:35Trade-off controller design
- 12:58Real-world results
- 13:40Conclusion
related talks
17mBridging the Gap between QoE and QoS in Congestion Control: A Large-scale Mobile...
25mPCC Vivace: Online-Learning Congestion Control
13mAUTO: Adaptive Congestion Control Based on Multi-Objective Reinforcement Learning...
22mLIVE Q&A #3
17mTowards provably performant congestion control
28mSafe sharing of the network with eBPF
31mTuning Linux TCP for data-center networks
23mACK to the Future: TCP in 2025
