Mobile video traffic: Alleviating the capacity crunch
Hulu, the online service that lets customers access their favorite TV shows from their computer, just went mobile. The company recently announced that its Hulu Plus service, which comes with a $9.99 monthly subscription fee, will now allow users to stream TV shows over 3G networks to Apple's (NASDAQ:AAPL) iPad and iPhone.
This is just the latest example of high-quality video content migrating to mobile. The trend is forcing wireless carriers to turn to video optimization and compression technologies to manage data traffic from YouTube, Hulu and other video providers--an issue that is only going to get more urgent in the coming months and years.
According to Cisco's Visual Network Index report from February, mobile data traffic will double every year through 2014 worldwide, and in 2014 video is forecast to make up 66 percent of all mobile data traffic. In addition, research firm the Yankee Group reports that the number of global mobile video/TV users will jump from roughly 250 million in 2010 to around 450 million users in 2014.

"They [wireless carriers] are addressing this as they feel the pain, specifically with network management and optimization solutions from third parties," said Fritz Jordan, an analyst at ABI Research. "They're band-aiding 3G networks for services the networks weren't even created for."
A perfect storm
Video on mobile is nothing new, of course. Sprint Nextel (NYSE:S) was an early pioneer of mobile TV with its GoTV service, and carriers have for years promoted their own branded mobile video services, like Verizon Wireless' (NYSE:VZ) V Cast, Sprint TV and AT&T Mobility's Mobile Video service (formerly called CV).
What has changed over time is the rate of innovation in devices, the advances of mobile broadband networks and the proliferation of high-quality video content across the mobile Web.
"It's a perfect storm of three dimensions," said Marc Zionts, the CEO of Ortiva, a vendor that specializes in video optimization. "It definitely intersects the movement from a wired world to a wireless world."
Carriers seem to understand the challenge, and carrier representatives acknowledged they will have to dig into their arsenal of tools and solutions to continue to deliver high-quality video offerings amid dramatic increases in demand for data.
"Delivering video over wireless networks is one of the most challenging issues we face," said Colson Hillier, executive director of mobile video products at Verizon Wireless. "There's multiple places the experience can be impacted, either positively or negatively."
Hillier said Verizon expects video will become a "material" part of traffic on its network, and that within 24 months it could account for the majority of traffic.
Mathew Oommen, Sprint's vice president for technology development and strategy, said the growth of mobile video traffic presents both a challenge and an opportunity. He said Sprint does not necessarily look at optimizing video as a way to deliver network efficiencies and capital savings--though that is obviously a large part of Sprint's motivation.
"We look at it as a way to enhance quality of experience, and through that process enhancing effective network utilization," he said.

Terry Cudmore, an analyst at the Yankee Group, succinctly summed up the Catch-22 operators face with mobile video: "Carriers are trying to get synched up to these great devices that are going to give the customers a great experience. But at the same time that user experience is what's clogging up the network."
Tools in the toolbox
There are numerous methods carriers can use to tackle mobile video and deliver high-quality, low-bandwidth experiences. Each solution, though, has its own complexities and, as many vendors and analysts noted, there is no single solution that effectively deals with the issue.
"Most operators need to look at it in a holistic fashion," said Patrick Lopez, the CMO of Vantrix, another video optimization company. "They need to put in place a number of mechanisms. The problem is that every month that they are not doing something about it, the problem continues to increase."
Here is a quick breakdown of some of the solutions carriers are employing:
| Adaptive streaming | The majority of mobile video traffic--67 percent, according to Vantrix’s Lopez--is accomplished via HTTP progressive streaming or download. This basically means that once a request for a video is made, the user can begin playing the clip before the download has completed. This approach is simple, but not very efficient for the network.
Adaptive streaming, by contrast, is essentially a handshake between the content and the device that has a mechanism to determine both the quality and speed of the network. In theory, this approach--which is not yet widely used--allows the device to better align a video presentation to actual network conditions. “We are expecting a strategic shift in terms of how this content is going to be encoded, both by the content owners and how it is going to be rendered by clients,” said Sprint’s Oommen. “It’s win-win for the customer as well as win-win for the network.” The problem with adaptive streaming is that there is no single defined standard for the technology: Adobe, Apple and Microsoft each have their own approach. “Even with adaptive streaming, the devices are not adjusting bandwidth conditions in real time,” said Ram Venketaramani, a product marketing manager at Openwave, which specializes in data analytics and optimization. “They’re always lagging the actual network conditions.” |
| Bit-rate throttling | One solution that Vantrix is pushing is bit-rate throttling, which essentially downloads video proportionally to the video encoding. So, for example, a five-minute video encoded at 1 Mbps can take about a minute to download. Lopez said that, sometimes, more than 50 percent of users will stop watching the video within first minute, meaning the network will have downloaded the full five-minute video and fully 80 percent of that video was not watched. Thus, the technique reduces network traffic by not delivering content that will not be watched by the subscriber. The drawback? It works in tandem with the more network-intensive HTTP streaming. |
| Transcoding and optimization techniques | There are numerous optimization, transcoding, transrating and compression methods. One, which Vantrix employs, re-encodes video proportionally to the bit rate of the download, which Lopez said “can actually guarantee you’re going to increase the user experience without the buffering.” Another method that Openwave uses is called a constant quantization algorithm, which, as it compresses video, increases how much spatial detail is saved for high-motion video frames while decreasing it for low-motion video frames. The upshot? “Even after compression, the end-user will have a very, very good quality of experience,” Openwave’s Venketaramani said. |
| Smart caching | Still another tool is smart caching, which basically allows for the storage of the source media as well for storage of transcoded versions of the video. This tool pre-fetches and buffers a video at the edge of the network and allows users to play the video when they establish a network connection. Essentially, the video is already queued up for the user, and saves on backhaul and radio-access network costs. "Hierarchical caching structure is going to be a key ingredient going forward for mobile networks," Venketaramani predicted. |
Each of these approaches is intended to address a problem clearly in the making: When Verizon launched the Motorola (NYSE:MOT) Droid X in June, the carrier highlighted the phone's video capabilities as well as a content deal allowing users to rent and buy movies from Blockbuster via the gadget. Carriers need to use these tools, Verizon's Hillier said, to make sure the "user experience matches expectations associated with these devices."
The bigger picture
Of course, video compression and optimization is just one part of a larger puzzle carriers must sort through as they figure out their network management policies.
"The experience of the consumer is so important, because the carrier is held accountable for the experience whether it's their problem or not, and they have to provide the capacity," Ortiva's Zionts said. "It's a dual-edged sword."
Carriers, especially Sprint and Verizon, have touted their move toward next-generation networks as a key advantage. Both mobile WiMAX and LTE will deliver higher throughputs, cheaper cost-per-bits and lower latency, the carriers argue, all of which would seem to augur well for handling mobile video traffic. But the carriers themselves acknowledge wider pipes alone aren't sufficient to address the issue.
"These are all tools in the toolkit," Sprint's Oommen said. "Just going to 4G is only one component. You need optimization, compression, edge service, caching. It is not either-or."
Verizon's Hillier agreed. He said Verizon is already thinking about the move to HD video for mobile devices, and concurrent voice and data streams. "Those things make the delivery more complex, vs. less," he said.
Jordan, of ABI Research, said there is no one-size-fits-all solution to handling rising data traffic, because each carrier has a different approach and assets. AT&T (NYSE:T), for example, has an extensive WiFi network that it can use to offload traffic, while Sprint does not. Still, he said, the problem remains. "If you're delivering something that is less than a quality experience, it's churn," he said.
"They're aware of what's coming," Yankee Group's Cudmore added. "Whether or not they're going find the absolute best way to deal with this--that remains to be seen."
Comments
One solution that has not been mentioned is delivering mobile video over broadcast networks. This would mean that devices would need a specific receiver and this would probably require a long conversion period within the natural device upgrade cycle of customers. Broadcast delivery is "one to many" and you could have unlimited viewers within an area versus a cellular unicast delivery of "one to one"-- so mobile broadcast would solve the capacity issues of mobile video consumption to keep the network free for more lucrative voice and other data applications. Assuming the mobile operator had a license and spectrum for mobile boradcast, they could build their own broadcast network. Or they could lease it from a network operator. One strategy mobile operators could use to monetize this network investment, following the Pareto principle, is to identify the top 5-10% mobile video users in their customer base and give them a free upgrade to a new smartphone with a mobile broadcast receiver (could be DVB-H, ISDB-T, TMB-- whichever is chosen by the operator as a broadcast transport standard). These 5-10% top users usually generate 60-80% of the traffic, so moving them to the broadcast network will immediately save on cost per bit. As you can see there are several factors to succesfully deliver mobile video over broadcast networks, but if market players can come together, it could be a win-win set-up for everyone.
The challenge with mobile broadcast is not radio device penetration, but rather that most mobile video is expected to be a personalized time-shifted/place-shifted experience - not a broadcast experience where all join in tonight at 8 to watch the latest issue of 24. So while broadcast is an end-to-end bandwidth saver relative to unicast technically, it is not a big saver in practical terms. Things like the Olympics or World Cup lend themselves to broadcast, but not YouTube or Hulu - those will be unicast / video on demand plays. The majority of mobile video consumption will be the latter.
Folks
I'm picking up with a link Marc Zionts was CEO at Westell a decade ago when they had hopes to be the DSL leader worldwide. His current company, Ortiva Wireless, was featured in an article at Fierce Wireless http://bit.ly/aCBUz6 on Mobile video traffic: Alleviating the capacity crunch. The article reports Mathew Oommen of Sprint and Colson Hillier of Verizon are looking at ways to reduce the traffic load as more people watch video. It appears that optimized streaming to mobile can have a major impact on network capacity.
While optimized video streaming sure sounds nice I much prefer forcing a progressive download model where the quality of the content does not suffer. On all these new devices poor quality of content is going to be significantly more visible than on devices even 2 generations old.
Ideally users could choose quality or delivery speed and the network would deliver accordingly.
Bit Rate throttling for Progressive Download and adaptive streaming for streaming delivery.


SHARE
WITH:
Comments (4) | Post a comment