What would have seemed a dream list of capabilities for high-speed modem and data support systems just six months ago is now within reach for commercial rollouts this year.

"We're much further along then I'd hoped to be at this point in the process," says Richard Green, president of Cable Television Laboratories, in reference to the industry's push for agreement on protocols. "It's looking like we'll probably make our mid-April goal for most interfaces."

This is saying a lot, given the high levels of functionality cable operators are seeking as they decide which media access control and physical layer protocols to tap for deployment. "The CableLabs process has forced things into the open, which means the best ideas from the pool of ideas are getting packaged into final products at warp speed," notes a leading manufacturing executive, asking not to be named.

"The technology is maturing very, very fast," agrees Mario Vecchi, who, as senior vice president of network technology for Time Warner Cable, is among those spearheading the protocol process. As a result, he adds, Time Warner will be able to launch major market rollouts of high-speed data service in systems passing anywhere from 750,000 to more than 1.2 million households this year with some very advanced capabilities built in from the outset.

"I can't be exact (about household count), because we're not sure about the magnitude of our internal ability to deploy these services, and we're also not sure about the ability of our vendors to deliver equipment in the quantities and at the quality of performance we require," Vecchi says. "But I can say (commercial deployment) will be in several large metropolitan areas."

In part, the increasing functionality available to the cable industry relies on the ever-growing demand from other industry sectors for advances in the network traffic management capabilities of asynchronous transfer mode technology. Equally important, cable's demand level is sufficient to push vendors out ahead of or around the ATM curve to add functionalities and efficiencies that aren't part of the larger enterprise and telecom environments.

At this point, it's getting hard to say what is ATM and what isn't, given the diverging cluster of industry-specific capabilities that have come under that moniker. From a cable perspective, the issue comes down to whether industry cost and performance requirements are best served through following the standards-driven cost curve or through proprietary approaches which make use of the ATM capabilities in the network backbone without rigidly adhering to ATM specs in the distribution link.

"Will all these (functionalities) converge on ATM?" asks David Fellows, senior vice president of engineering and technology at Continental Cablevision. "In the cable industry, we don't have to wait for the perfect (ATM) protocol. We can launch services on carriersĐwhich means we can launch converged services without having to wait for convergence."

As a case in point, he notes that ATM, designed to meet the dedicated, star/star networking needs of the telephone companies, accommodates capabilities which cable doesn't have to worry about. "If we want to deliver a CD multimedia service, we don't have to worry about things like 41 kHz sampling rates and 64 kilobit sampling for N," he says, in reference to the time division techniques of digital telephony.

"People will probably use different vendor-specific MACs at first," Vecchi notes. "This will likely be standardized in the future, but it's not necessary now."

ATM in the backbone

At the backbone level, ATM is well-equipped to deliver what cable needs, though Vecchi points out, "we're requiring vendors to make their utmost efforts to make signal conversion as cheap as possible." But new ATM capabilities are positioned to support integration of centralized functions covering data and the rest of cable operations, including network management and evolution of operations services; local caching of content; multiple security and billing interfaces; and, of course, connection through the cable gateway to data application gateways in the local community and worldwide.

No one is closer to efforts to put ATM to work in the cable data backbone than Milo Medin, vice president for networks at @Home Networks, who engineered the ATM backbone for NASA. "The Internet is at the point where growth is taxing the ability to handle the traffic," Medin says.

"Because we have architected our backbone features for efficiency and scalability from Day 1, I think we can get around some of the more well-known problems people are having."

Medin says @Home is negotiating with potential carriers for transport connections between the service's ATM switches and regional cable data centers. The broadband switches will direct high-speed traffic over the backbone, allowing @Home to avoid dependence on any one interexchange carrier, he adds.

As one way to speed cable customers' access to the Internet, the company is installing servers with large memory banks to support caching of material downloaded to the local site by end users. In this way, Medin says, anything recently tapped from the Internet by a local customer will be stored for a period of time, allowing other users seeking the same material to get it quickly rather than waiting for it to come through the long-haul pipes to the local server.

"The Internet today depends on relatively slow-speed interexchange lines connecting servers," Medin says. "Caching is a pretty low-cost way to help get around that."

Four gigabytes of storage costs about $900, he notes, adding that "in six months, it will be half that much."

Another innovation aimed at getting around the long-haul speed delay is multicasting, where information in high demand and in need of constant updating is put out continuously to all points over the national backbone @Home is building.

Distribution side requirements

Whether they are part of the official ATM process or cable-specific spinoffs, advanced data management algorithms are also working wonders for the local distribution side of the data network. Vecchi's list of doable tasks includes:

  • Link level encryption - "This provides protection from piracy and is important to management control," he says.
  • Common spectrum management interface - "This allows the network operator to allocate spectrum, to manage the bandwidth allocation to accommodate a variety of heterogeneous technologies."
  • User account management - This suite of capabilities includes a login validation, activating link-level encryption and assigning IP addresses and enabling the TCP/IP stack; manages subscriber accounts through features such as passwords, changeable access levels and parental control; and provides for shifting user accounts to new location and for user dial-in access to home-based modems.
  • Enterprise connectivity - The system must support a class of third-party premium information services, such as telecommuting, education, remote medicine, etc., all launched from the basic service interface but provided and billed separately.
  • The cable modem with self-installation and adaptive behavior - The modem has to adapt in a seamless way to the network, mapping the MAC onto the PC and registering the TCP address without operator intervention or any downloading of software.

This last requirement is vital to low-cost operations and "one of the most difficult to resolve," Vecchi says. "I think the solution we have arrived at is that the cable installer splits the cable to where the user wants to connect the modem. He carries a portable computer with an interface card and enough processing capacity to ensure the connection supports sustained maximum peak rates." The modem does the rest.

ATM in distribution

These kinds of capabilities, supporting a potentially vast array of links to a great variety of multimedia packages, are now moving out of R&D into the production line, in many instances employing refined versions of new ATM capabilities devised for end-to-end connections in the business community. In particular, there's growing support for adapting a new desktop ATM access format to cable's requirements.

"We absolutely expect that ATM will be one of the standards adopted for high-speed data service to the home," says David Nelsen, director of the outbound technology group at FORE Systems Inc. FORE, a leading manufacturer of ATM switches, is now teamed with both Nortel and General Instrument Corp. in separate development projects aimed at delivering end-to-end ATM connectivity to households by year's end.

The 25 Mbps ATM/PC interface has long been supported by IBM, FORE and a handful of others as a "low-speed" version of ATM that could be delivered over standard gauge twisted pair LAN wiring. With backing from industry standards bodies, the concept has moved into production, with FORE and others planning delivery of ATM network interface cards for PCs and work stations by the beginning of next year.

The FORE technology under development for cable networks is a stripped-down version of 25 Mbps ATM that will be available much sooner, officials say. Whereas the ATM network interface cards for LANs support a broad range of protocols common to the business environment, the cable version will focus on data packaged for transport in the IP format, which includes all on-line as well as Internet services and on-line stored media such as CD-ROMs.

These cards will be priced at about $200 and will be put to use in the cable modem-to-PC interface of the GI/FORE product, eliminating the 10 Mbps 10baseT "bottleneck," says Ed Zylka, director of marketing, network telecommunications systems at GI Communications, who declines to say whether there will be a 10baseT PC/modem interface option offered with the ATM product line. "As a company, we're saying ATM is going to play a strong role," Zylka says.

"This doesn't mean we're coming out and saying that the PC connection should be based on ATM or that all set-tops are going to migrate to ATM," he adds. "But it does say that the next step for GI beyond the development we've done for digital TV and cable-delivered data is to create a strong core network technology."

The GI strategy drew praise from Microsoft Corp., which has been working with GI and FORE for the better part of two years and has long espoused ATM as the best framework for all types of digital communications in the future, including MPEG video. "We believe this announcement will help speed Microsoft's efforts to bring new services to PCs and televisions," says Microsoft senior vice president Craig Mundie.

The importance of content

Operators stress the importance of their foray into data as much more than Internet access, which places further demands on the system protocols. "We're not in this to become another Internet access provider," Vecchi says. "Our research shows that customers want much more."

For example, operators and commercial on-line content providers are looking at ways to exploit CD-ROM content, including full-motion video components, as well as to create new content built around the widely used authoring tools of the CD environment. "We see a lot of work going on in the commercial on-line community that hasn't been made public that will take advantage of these high speeds," says Vecchi.

In another case in point, Continental, now in a 200-home trial in preparation for commercial rollout in the second half, is working on content as well as technical details as it moves to widescale deployment, notes Jeff DeLorne, executive vice president for the MSO. He says Steve Hill, the MSO's new senior vice president for high-speed data, is focusing his energies on working with a variety of entities to put together locally- as well as nationally-oriented content, including services with a large video component.

In another major development aimed at fostering more advanced content, @Home has tapped Macromedia Inc. to supply software tools that could speed development of multimedia content uniquely suited to the high-speed data access environment. The two companies say they will put Macromedia's Shockwave software to use in conjunction with other software from Macromedia, Netscape Communications Inc. and Sun Microsystems to provide a user-friendly environment for creating and linking multimedia across the World Wide Web.

"The economies of scale we achieve through our distribution agreement with TCI allows us to negotiate from strength with companies like Netscape and Macromedia as well as hardware suppliers," says @Home's Medin. "We can be the low-cost provider for doing things which it would be hard for individual cable companies to do on their own."

@Home demonstrated some of the combined capabilities of these software systems at the Western Show with applications such as multimedia game playing and video clips imbedded in the network's home page. Shockwave works with Macromedia's Director, the multimedia industry-standard authoring tool, making it as easy to tailor content for distribution over the World Wide Web as it is to prepare it for other distribution formats such as CD-ROM, says Jane Chuey, spokesperson for Macromedia.

"We use the same intuitive approach that we use with Director, which means you don't have to be a software programmer to port material to the Internet," she says. Sun's Java, which is a language that does require a programmer's skill, takes the content development another step, allowing developers to add interactivity within any downloaded page. The combined power of these tools will make fast action multimedia applications readily available to anyone with a cable modem, officials said.

@Home and Macromedia also say they were cooperating on development of new tools that will go beyond the static text and graphics capabilities of today's HTML (hypertext markup language). The tools will allow developers to use the latest video, sound, 3D and interactivity support systems in conjunction with rich multimedia effects.

One aspect of this initiative uses the principles of a new software system developed at the University of Colorado, known as "Harvest," Medin notes. "This is a new kind of search engine that includes modules that go beyond what can be done with HTML," he says. "For example, it can be used to search and retrieve something from an MPEG file."

FORE's Nelsen argues that Ethernet "is not particularly useful for audio and video," whereas the ATM system, which will deliver service at minimum speeds of 2 or 3 Mbps to an individual user during peak usage, can be set to assure the minimum throughput and continuous bit stream is maintained to support video. But Medin argues that it is too soon to take ATM all the way to the home, and others contend that, even if the ATM package is taken to the modem, it should be converted to 10baseT for low-cost insertion at the PC.

Medin, Vecchi and others banking on the 10baseT interface make clear they want the kinds of capabilities Nelsen alludes to built into their systems. In other words, whether it's called ATM or something else, cable is seeking a true multimedia data system capability that looks beyond today's Internet and commercial on-line service base. By all appearances, it's going to have it.