I recently had the opportunity to participate in the Q4 Streaming Video Alliance (SVA) meeting in Lisbon, where, among other things, I had the pleasure of presenting on the topic of ways to improve Quality of Experience (QoE) for streaming video.
In my talk, I explained how the “last mile,” that stretch of network from the edge node of the CDN to the video player, often looks like a busy highway at rush hour where several vehicles have already broken down. In other words, streaming video has a lot of competition to get to the end user’s mobile phone, TV or other device, and most of it is out of the control of the provider, the end-customer and the CDN.
Congestion control is key, and the congestion control mechanism in TCP -- the Transmission Control Protocol through which streaming video is sent over the internet -- is ancient, having had minimal improvements made to it in the last two decades.
There have, of course, been attempts to create conceptually new congestion control algorithms to replace or augment TCP, and I talked about two of them, both open source:
Bottleneck Bandwidth and Round-trip propagation time or BBR, created by Google, and
Performance Oriented Congestion Control or PCC, created by the team of myself under the auspices of the Hebrew University and my colleague Brighten Godfrey, of the University of Illinois at Urbana-Champaign.
These two approaches come from radically different philosophies, with significantly different effect on the actual QoE.
I’ll let you see for yourself in the video below how PCC generates significantly better QoE results based on every relevant KPI (startup time, resolution, etc.) than the traditional TCP. If you watch to the end, you’ll also learn about my vision for using the PCC framework to “customize” the congestion control parameters depending on the type of application people are using and the prevailing network conditions.
You can view the slides here.
I hope you enjoy the presentation, and if I can answer any questions about it, please feel free to contact us.