Even while live video is catching on as a popular way for people to have their calls and meetings, we’re starting to see carrier plans that put caps on total data use over the networks. As one recent example, AT&T has announced DSL and U-verse monthly limits of 250 Gigabytes and less, and linking any gadgets to AT&T’s new wireless plans will hold them to 2 GB per month for the higher tier, and a quarter GB (250 MB) for the cheapest rate.
Those carriers are implementing these limits, they say, to keep a few voracious movie-sucking data vampires from pulling the network down around our heads. Whatever the merits of this argument (they don’t actually say “vampires”), the fact remains that the rest of us want to use video, too. How is vampire video different from ours? How do these carrier limits affect the ability of the rest of us to use video for our live conversations and meetings?
To answer these questions, let’s take a look at how videophones, telepresence, and videoconferencing use data.
The continuing trend in video communications is an increase in the quality-to-data ratio. I’ve followed the progress in lossy codecs over the past few decades – lossy codecs are the ones that don’t insist on zero errors, only invisible ones; video communicators tend to be a fairly pragmatic bunch so video communications systems rarely use lossless codecs, What I’ve found is that image quality has been preserved while the data rate has steadily decreased. Stated conversely, image quality has increased for the same data rate. So for example, where a low-resolution CIF image (352 x 240 pixels) fully topped up a T1 line of 1.6 Mbps in 1993 with MPEG1, today HD video (1920 x 1080 pixels) happily fits in the same bandwidth using the H.264 High Profile standard, an overall improvement of 25 times.
In real use, video communication systems typically run at rates from 256kbps for a tablet-size display, to streams of 2-4 Mbps for 1080p HD connections. Again, these numbers continue to improve; the open standard High Profile for H.264 yields another two times improvement, for example, so that a full HD video link can nicely fit in a megabit.
One difference between a video call and a movie download is that streaming video is one-way, while face-to-face video is symmetrical – similar data rates go both ways. At the moment, though, data caps are only being applied to download, so this isn’t an issue.
We can make some more assumptions. Most connections won’t be running 24/7. If we assume an average daily usage of four hours, then a Samsung Galaxy or Motorola Xoom running normal video (H.264 main profile) at 256kbps will top out at about 11Gbits/month. A single-screen HD videoconference draws 86 Gbits/month, and a full three-screen immersive telepresence will triple this, at about 260 Gbits/month.
If you’re nervously glancing back at the plan limits, don’t panic yet (well, you can panic at the smaller wireless plan, 250 milligigabytes, but unless you live a life devoid of all imagery, you’re probably not choosing that one anyway). Remember, those plans are in GB, while video rates are commonly stated in Gb. Bits, not bytes. Converting video usage to the same units, that wireless Android pad will draw 1.4 Gbyte/month, leaving about one-third of the wireless plan’s 2 Gbyte/month allocation (and four hours a day is probably a bleary-eyed, aching-wrist extreme for the pad conferencing scenario over a cellular network). The wired connections are even less fettered: in an extreme case, with full immersive telepresence, three screens at four hours a day, we would account for only 33 Gbytes through the month, one-fifth of the cheaper plan’s allowance.
AT&T says that average usage is 18Gbyte/month and that less than two percent of their customers will be affected by the wired limits. Whether two percent is the right number or not, even frequent video users still look pretty safe. So go ahead: chatter away, make faces, and show us all your latest moves. Even in this new age of usage limits, video communications still fit easily into our lifestyle.