Reporter with Silicon Hills News

Jeff Andrews, Past-Director, Wireless Networking and Communications Group at the University of Texas

An Exabyte is a quintillion bytes…or a billion, billion bytes. Every month the world generates 1.3 exabytes of data in the mobile space.
“Last year’s mobile data traffic was eight times the size of the entire global Internet in 2000,” Cisco reports.
By the year 2016, 70 percent of that will be video, according to Cisco.
The industry is wondering how it’s going to keep up with all that data. And how it’s going to monetize it when so much of it is free. Those were some topics of a panel Friday at Austin Technology Incubator’s Texas Wireless Summit Friday at the AT&T Executive Education and Conference Center. Jeff Andrews, past-director, Wireless Networking and Communications Group at the University of Texas moderated the panel which included Jeffrey Foerster of Intel, Felix Fernandez of Samsung Dallas Technology Lab, Arvind Raghavan of AT&T Labs and Mark Newbury of Huawei.
Video is in some ways a forgiving medium. The quality depends largely on the perception of the person watching. So the industry extracts the opinions of many people to identify a quality standard it can use in algorithms to recreate that level of quality. Video quality also depends on what device people use to look at it. A close-up of something small requires fewer bits to look sharp than a football play in a stadium. But even a football play video that uses few bits can look good on a mobile screen.
The video that concerns the industry is all the real-time video. Streaming, personal videos people make and share of their lives—travelogues in real time, for example. With the increased burden on the network, latency is a big problem for those kind of videos. Interactive videos also need super low latency rates, especially video teleconferencing.
“Quality is only one aspect of the experience,” said Foerster. “Rebuffering is a huge issue that has the biggest impact on experience. Quality variations over time, audiovisual sync, startup delay, all those things come into play making this a rich space for optimization.”
Fernandez explained that a new video recording standard, which will probably be named HEVC, will be finalized in January 2013 and will significantly increase coding efficiency and compression, allowing higher quality videos to be transmitted at lower bit rates. Part of his work is in green video, which allows an encoder to send the video with information that would allow a receiver to decode the video at a 25 percent savings in power level. The end user would have to choose to trade off some quality for energy savings.
But, as Fernandez pointed out “Video streaming is very power hungry application. With social streaming you’re walking around a new place, you capture a video and upload it to the cloud so your friends and family can watch it somewhere else.”
The more that kind of video exchange grows, the more willing, he said, people will be to trade off a bit of quality for lower power consumption.
Part of the industry’s job will be classifying and directing traffic through the different mediums according to Raghavan.
“How do we manage the layers within cellular?” he said. “The carriers have several different dimensions with a reasonably good quality of experience across a wide base. How do you optimize across these dimensions and user profiles?”
Stored video, he said, can be transferred in fairly large chunks whereas as you move down the spectrum toward live and interactive you start to get more latency issues and the packets get smaller. They need to be able to groom the traffic coming in to make the radio network more efficient.
Another panelist, Newbury introduced the idea of monetization with an illustration of “black hole of consumer demand and provider investment.”
“The network itself is pouring investment dollars into this black hole. The revenue per bit for video is orders of magnitude less than the revenue per bit of SMS.”
In order to come up with better quality and revenue models, he said, the industry needs better ways to measure activity in the same way that the wireless industry measured dropped calls against load volumes.
Current revenue plans, he said, are bent not so much on raising revenue as on not breaking the network. Current pricing plans are just to protect the network. But it is too early, the panel decided, to answer the question of “Who is going to pay for this?” If the content in question is a professional sporting event, for example, is it the team? The network or the end user?
Raghavan, as the only provider representative, said models he has seen show that more data on the network would definitely result in new revenues.
“If you have a $30 a month plan and have a 3 gig bucket, if you consume more, the plan is going to go up. But we are also looking at more subtle and interesting ways of developing income.”
And that was all he would say on that….