evising solutions to effectively handle high-bandwidth video--streamed or downloaded over wired or wireless networks-is one of the big design challenges facing system engineers working with consumer terminals and mobile phones, as well as architects of network infrastructure.

As consumers hunger for richer media content and service providers eye bigger revenue streams from such demand, any intervening technology hang-up takes on the proportions of a crisis.

A typical hang-up is the quality of high-bandwidth video streaming over the Internet. Today, this is severely limited by a host of factors, including the bandwidth available in the backbone, the speed of the connection, latencies caused by multiple hops between video origination sites and end-users, and packet losses caused by Internet congestion.

As Mahal Mohan, director of product marketing at Digital Fountain Inc. (San Francisco), put it, "The Internet simply isn't designed for sending rich media to a lot of users."

With limited bandwidth in today's cellular network infrastructure and a network more prone to packet loss, sending and receiving good quality video using one's handset remains a fairy tale to most mobile users.

Even if enough bandwidth becomes available in the emerging 2.5G and 3G networks, those who design video-capable mobile units must deal with such looming technical issues as processing power on a handset to decode and encode video, battery life and display quality.

Further, there is confusion in the market over a diverging array of formats, platforms, devices and connections.

For example, the streaming-media market today, rather than solidifying on one standard format, is growing ever more fragmented. Formats are increasing. Solutions to this logjam get more complicated when content owners start to think about spreading their high-bandwidth video to a broader range of receivers-PDAs, set-tops, wireless devices, etc.

Unlike a personal computer, where almost unlimited processing power makes it possible to decode high-bandwidth video compressed in almost any format, most non-PC consumer devices (mobile phones in particular), cannot afford that luxury. If content owners want to reach the largest possible audience-whether they are PC users, TV viewers or mobile phone/PDA users--they may have to depend on emerging networks that come with some level of built-in intelligence. Ideally, such intelligent networks should be able to transcode video formats and optimize the image quality of high-bandwidth video in accordance with the available processing power or display resolution on a receiving device, without wasting the overall network bandwidth.

Wireless solution
A host of companies today are attacking the core issues of handling high-bandwidth video over the wired and wireless Internet. While their approaches and focuses differ, many of them are struggling to strike a balance between roles the receivers and network infrastructures should play.

Companies such as PacketVideo Corp. (San Diego) are leading the pack, with the temerity to venture into the brave new world of delivering video over the wireless network. Indeed, although a number of companies-including Real Networks Inc. (Seattle) and Microsoft Corp. (Redmond, Wash.)-already offer streaming audio/video technologies for the Internet, few technology companies have focused on tailoring their technologies for the wireless market.

"Nobody has built a streaming solution as comprehensive as ours, optimized for wireless networks--where channels can be easily lost, and there are huge fluctuations in available data rate," claimed James Brailean, president and chief technology officer at PacketVideo.

Underlying technologies-now offered by PacketVideo to chip companies, handset vendors, wireless carriers and content owners-include embedded software for mobile devices, technologies for serving multimedia content over the wireless network, and authoring tools to help build wireless multimedia applications.

One key technology developed by PacketVideo and put in place in the multimedia wireless access network is the company-designed wireless edge server. Based on MPEG-4 temporal scalability and PacketVideo's own dynamic rate control technology called FrameTrack, PacketVideo's wireless edge server dynamically manages highly variable wireless bandwidth for streaming applications. The PacketVideo Player software on the handheld detects common fluctuations in bandwidth and communicates back to the PacketVideo server, which in turn regulates the frame rate. FrameTrack, in essence, provides enhanced bandwidth management, including scaling to support data rates from 9.6 kbits per second to more than 384 kbits/s. This means one encoded MPEG-4 file can deliver multiple streams at multiple bit rates to multiple devices.

At a time when no wireless video infrastructure exists in the market, Brailean acknowledged, "We had to develop an end-to-end solution," ranging from software that goes into handsets to network optimization solutions for servers in the network and encoding technologies for content applications. PacketVideo needed to do that "just to get the market going," he said.

Meanwhile, some in the industry call wireless video downright impractical, and remain skeptical of its commercial viability. "Consumers are not paying for video even over the wired network," observed Philippe Kahn, chief executive officer of LightSurf Technologies Inc. (Santa Cruz, Calif.). "Video makes a cool demo. But nobody wants to pay for it. Video will not offer a kind of return on investment carriers are looking for."

Rather than going after the wireless video market, LightSurf is in its fourth year of developing an advanced wireless platform for delivering digital still images. The plat-

form is optimized for sending and receiving digital photos from any wireless device-cellular phone, PDA and digital camera--to any other wireless or wire line device. The LightSurf platform consists of a scalable suite consisting of a tightly coupled server, e-commerce and wireless device components designed to deliver instant wireless digital photography, according to Kahn.

"Anybody can put together a quick demo of sending a JPEG-compressed photo over a wireless network to another wireless device," said Kahn. "What's hard is to make these massively scalable and economically viable for millions of users to work with, and for carriers to create a real incremental business."

The claim-to-fame feature on LightSurf's platform is its ability to capture and wirelessly transmit JPEG2000-compliant files at two to three times the speed of existing technologies. "This is extremely crucial, because being able to send pictures faster means a cheaper transmission cost for users while it also allows carriers to handle a three-times bigger load," Kahn said.

LightSurf does it by combining very lightweight device applications on handhelds with a LightSurf Media Gateway that provides the server side of applications. More specifically, Media Gateway enables media acceleration as well as so-called "smart-connection recovery."

Kahn explained that the company-developed smart-connection recovery effectively optimizes file transfer when cellular calls are dropped. "It remembers the last point where the call was dropped and recovers the connection from there, rather than sending the whole file all over again." He painted a possible scenario in which 10,000 people in a football stadium happen to take a picture of a touchdown by using their mobile phones featuring a digital camera, and decide to send it to friends and families simultaneously. Without a proper network infrastructure to handle peak load, "Everything can go haywire," Kahn said. "It's extremely critical to build a platform that can provide a viable, profitable business. This is not just for a simple science fair project."

As with others developing solutions for Internet-based digital media delivery, Kahn is a big believer in an intelligent network. "Our vision of the future is that the network will carry more and more load," he said. While a device should offer a radically simple user interface for easy point-and-shoot and dispatching of pictures, the intelligence in the network-such as LightSurf's Media Exchange server software in the network-should be able to enhance pictures, reformat images and optimize their quality to a receiver's display, he said.

Adding intelligence
While companies such as PacketVideo and LightSurf are focused on end-to-end solutions, a few companies are homing in on technologies that add intelligence directly to the network itself.

Generic Media Inc. (Palo Alto, Calif.), for example, will offer what the company calls the Transformation Engine. Sitting at the core of the Generic Media Publishing Service, the engine is designed to process, optimize and deliver digital content on the fly and on demand.

Angela Lai, vice president of engineering at Generic Media, maintains there's "not a chance" for the industry to agree on a single video format for video distribution over the Internet. Not now, and most likely not in the future, she said.

Formats confusion
Today, content owners face confusion over picking the right types of streaming formats, sending at an appropriate bit rate, and considering a variety of distribution networks and a broad range of receivers that decode and display images.

Generic Media believes its job is "to take that decision out of media companies and let them focus on what they do best," said Lai. Simply put, the Generic Media Delivery Manager software detects the user's software, hardware and bandwidth environment and delivers the optimal stream. Generic Media streams require no additional plug-ins, bandwidth selections or any other action on the part of the user.

"Content providers will supply us with only one file-the master asset-and we manage it effectively by deferring the processing of content until it is requested by a user and for a specific media player and connection type," Lai said. "We see ourselves as a new component to content-delivery network vendors such as Akamai or Digital Islands." Lai denied that Generic Media's transcoding service itself may become obsolete as more processing power is added to a client so that a receiver can handle all the necessary transcoding. "Non-PC devices such as PDAs or cell phones will neither have enough processing power nor enough memory space to store streaming media. How do you upgrade your phone?"

Clearly, new technology companies set to solve broadband video delivery over the Internet today are no longer looking at such a brute-force approach as throwing more and more servers at the edge of the network.

While pioneering companies such as Akamai, Digital Island or iBeam have already tried to solve the problem by bypassing the public Internet with caching and streaming-edge servers, many in the industry today warn that such content-delivery network solutions are not designed to scale for broadband applications.

"Many people underestimate the sheer size of video," said Satish Menon, chief technology officer at Kasenna Inc. (Mountain View, Calif.). They don't allow enough bandwidth in the backbone, while they build edge servers that don't have enough storage capacity to handle

carrier- and commercial-grade streaming media, he said. "They have no idea how much space broadcast-quality video streams could take up." When today's edge servers are tasked to implement Internet ads with high-bandwidth video, Menon predicted, they won't be able to handle it.

For its part, Digital Fountain is angling to offer "a quantum leap in content delivery." The startup is proposing to transform content into "Meta-Content" packets. Meta-Content is essentially a mathematical equation.

Multiple packets
Meta-Content is made up of multiple Meta-Content packets, each of which represents elemental information from the entire body of content. While properties of each Meta-Content packet are unique, they can completely substitute for each other. Once content is sent to a Digital Fountain server, the file is scanned and stuffed into Meta-Content packets. The packets become available to all clients.

By turning the original content into Meta-Content packets, "We don't make it the quality of content worse," promised Mohan. "We'll let you replicate the original content." Basic coding theories used in Digital Fountain's Meta-Content packets are similar to those applied to forward error corrections.

The emerging technology still needs to be put in the network to prove its efficiency, but some analysts already hold high hopes for it.

"Digital Fountain's servers enable cost-effective, on-demand delivery of highly popular static or streaming files

to very large, concurrent audiences," said Michael Hoch, an analyst at Aberdeen Group (Boston).