Virtual Reality: Bandwidth Reality

Great thread Greg. Gaming is the low hanging fruit. Also Telepresence sans the dedicated room. Virtual shopping for goods and travel. Education & corp Training. Medical training (actual telemedicine more AR than VR). Produced entertainment is getting major investment. Zuckerberg's vision is to take the FB conversation into virtual rooms and god knows what else. Adult content. Live events and especially sports have major challenges for real time production, but have immense potential. Yes, most are unknowns still but consider that the population's appetite for it is being whetted with free or cheap smartphone VR holders, and although smartphone based VR has limited used (will make you queasy due to lag/jitter/resolution) it will get better in time. I personally don't believe this is the next 3G TV false start. I may be drinking the cool-aid but I demo'd the Vive at MWC and the experience is utterly transformational.

Cornel Ciocirlan One of the benefits of gigabit networks is lower latency. You could have your super computers centralized in CO's/HE's. Other than special forces training what apps require very low latency, high resolution and 360 FOV? Tele medicine or remote surgery? I hope the doc isn't looking behind during the operation :-) Guess it depends what your strategy is. High Volume or Niche low volume but high margin?

Cliff Raskind I'm focus on bits per second since that drives network decisions. I fail to see how 4K or even HD VR will be anything more than niche applications such as theme parks or special forces training for the next 7-10 years. Also, the human eye is Extremely forgiving (see JPEG and MPEG). The human ear is much less forgiving.

VR bandwidth will be higher because latency is critical as Charles indicated. The 720p 17Mbps assumes encoding / compression - and most YouTube content is heavily compressed. Heavy compression (for bitrates) and low latency are competing goals ... For interactive VR applications the level of compression will have to be much lower in order to avoid shipping supercomputers with each VR set. Lower compression means higher bandwidth. It may not be 25x but it's not going to be 2x either ... For "pre-encoded" (non interactive) content 2x may be the right multiplier but if interactivity is desired it may well be 10-15-25x.

If the reality you're virtualizing is cinema or TV (like the Avegant Corp.Glyph or Vuzix Corporation's new iWear glasses) then the 2x bandwidth makes sense. Bandwidth and server peak demand planning should be based on "just in case" scenario's where the user might look up/down/left/right and have the imagery ready to go. When I did this kinda stuff for Nokia's Maps Platform we needed to have preloaded maptiles (or vector graphics data) for a seamless experience if the user swiped north/south/east/west. My colleagues in SA's networks team postulated that predictive technologies are prolly good enough that you don’t need to stream all 16 scenes at once (separately for each eye) on the off-chance that you quickly turn round 180 degrees, but based on the movement of your head the system may decide to begin delivering the stream one or two scenes to the right of current vision if you start moving your head in that direction – so going to be a trade-off between an experience that requires massive bandwidth and one that doesn’t introduce too much lag/jitter as you turn your head. We'd spitball it that if you have VGA – you are looking at 720p video that is probably closer to 1.5Mbps or ~ 11MBytes for a minute. At which point you’d be getting closer to 100MB for one minute when sending 4 scenes to each eye plus other info. Any more accurate info certainly welcomed! I'm not sure about full 4k but even HD screens produced the pixelated “Screen Door” effect, like the Oculus v1.0 suffered from. I know that in handset based VR like Sansung Gear VR this effect was still quite bad in the Samsung Galaxy Note 4, and was improved, though still discernable in the Galaxy S6. Haven’t tried the latest S7 based VR.