Performance of Low-Latency DASH/CMAF and Low-Latency HLS Systems
summary
Experimental systems use FFmpeg + Apple HLS tools for LL‑HLS, Node.js DASH and DASH.js variants for LL‑DASH, and Mahi‑Mahi traces from Verizon (1.3 Mbps avg) and T‑Mobile (1.6 Mbps avg) over 10 minute sessions. Profiles cap bitrate at 1.2 Mbps down to 279 kbps. Results show LL‑DASH often hits 3–7 second glass‑to‑glass latency, LL‑HLS issues many more HTTP requests, and some clients AVPlayer buffers once vs 60+ for others under handoff conditions.
AI-written summary · about
chapters (12)
- 0:00Low-latency streaming study intro
- 1:32Low-latency HLS/DASH evolution
- 3:39Streaming systems evaluation
- 8:42Verizon 4G LTE results
- 14:20T-Mobile 4G LTE results
- 15:45Key findings and limits
- 18:51Q&A on latency
- 20:16Measuring streaming latency
- 21:19Sub-segment latency strategies
- 23:05Near real-time streaming advances
- 25:08Future client algorithms
- 26:53Closing remarks
related talks
19mLow Latency Live from a Different Vantage Point
26mDASHing Into the Wild with VP9
24mMeet the Packets: How audio travels into your browser
36mCHUNKY MONKEY
35mLive Low Latency Video Streaming
43mModern Streaming Video for the Jamstack
44mLive Video in a Modern Web Environment
12mLHLS(.js): Why Hls.js is standardizing low-latency streaming