RBOC plans to deliver integrated services

Phone or cable network architecture may seem like an esoteric topic, but it is vitally important. The network architecture determines what it will cost to wire a community for high-speed, high-bandwidth service and the type (and quality) of service the network can provide.

The current telephone network is a “star/star/star” network: conversations are routed through a series of “star” hubs or switches all the way to the home. There is a separate pair of wires for each phone line coming into your home, and a different set of wires for the lines running into your neighbor’s house.

Most of the plans for upgrading the phone system envisage retaining this architecture, but running high-capacity fiber to the neighborhood star level Q or even all the way to a connection box on the side of the house. The existing copper wire would carry the signals the last bit of the way.

Cable television uses a “bus” architecture. A common coax cable serves all of the houses in a neighborhood. Everyone taps into the same wire.

Initially, the cable industry planned to replace all but the last mile of cable with fiber to create the much-talked-about 500-channel cable service. However, the vision has expanded during the past 18 months. The cable industry’s Cable Laboratories has appropriated the latest telephone ATM (asynchronous transfer mode) technology and married this to the coax cable run that serves a neighborhood. (For an explanation of ATM, see p. 14.)

The resulting “star/bus” network combines the advantages of a fully switched, two-way packet network with those of using shared coax cable rather than separate copper or fiber runs for the “last mile” to the home (and through the home). This was demonstrated for the first time in a live “CableNet” demonstration at the recent Western Show in Anaheim, CA (see Western Show story, p. 18).

In a remarkable turn of events this past month, Pacific Bell announced that it was returning the favor and stealing back CableLab’s hybrid star/bus technology. It plans to replace its existing copper phone wiring with a new system that looks exactly like those to be implemented by the cable operators!

— Pub.

The frenzied jockeying for market position among telephone and cable companies reached new levels of unintelligibility last month with Pacific Bell’s embrace of cable television networking technology as its ticket to success in the digital future.

The regional Bell operating company announced it will spend $4–5 billion beyond already scheduled outlays to replace copper lines serving 5 million California households with AT&T-supplied broadband “distribution plant” during the next seven years. The still-evolving network scheme begins with a commitment to marry Bellcore’s broadband telephony protocols to the “star/bus” network topology pioneered by the cable industry.

The project goes well beyond any use of cable-derived technology yet announced by a telco. Unlike US West, which is deploying a parallel network — a cable-like fiber/coaxial overlay — in conjunction with a fiber-to-the-curb conversion of its existing voice network, PacBell is integrating its voice services onto a single fiber/coaxial pipeline and eliminating the existing copper network altogether.

INFRASTRUCTURE COST ASSUMPTIONS OUT THE WINDOW

The PacBell decision adds a major element of uncertainty to the red-hot cable/telco debate. It challenges existing assumptions about timing, costs and regulations in deployment of the national information infrastructure. In effect, it suggests that the telephone industry’s long-standing message to regulators concerning the ultimate benefits of point-to-point, star/star/star fiber networks is wrong, and it threatens slower delivery and higher prices for digital communications.

Jumping on the star/bus bandwagon. The performance capabilities of the new star/bus design, as touted by a growing number of major telecommunications suppliers, extend to virtually any type of service envisioned for broadband networks. What PacBell is asserting is that these manufacturers are not only justified in their claims but also have the cheapest means for delivering such services.

PacBell’s projected costs come to about $900 per household, which is some $200 or more below the costs anticipated for other approaches. Moreover, PacBell’s $900 comprises the entire increment above the costs of already-scheduled digitization and fiberization of the core central-office backbone, whereas the broadband costs cited for other topologies exclude narrowband distribution elements, such as any fiber that might be cost-justified for voice service under various copper depreciation schemes.

Taking the wind out of cable’s sails. Under PacBell’s plan, high-speed data and interactive digital video circuits would be deployed to more than a third of the state’s households by the end of the decade (and within currently allowable spending parameters). This represents a market base of users that the cable industry would be hard-pressed to match in that period of time. Indeed, if PacBell were to garner government support for its agenda, it could take a lot of the wind out of cable’s sails. The cable industry’s growth strategy has been pegged to the cost and performance advantages of the same type of network that PacBell is proposing, but PacBell wants to build at lightning speed.

“We’re pursuing an ‘if we build it, they will come’ strategy with a twist,” says Ron Stowe, vice president, Washington operations, for Pacific Telesis Group, parent to PacBell. “Here, the ‘they’ is also our competitors, who are coming whether we build it or not.”

The trick is to provide protection against facilities competition, Stowe says, even if the competitors have an advantage with regard to regulatory freedoms. “If public policy doesn’t change and become more forward looking, we’d better be in a position of having such an impressive array of facilities available that the competitors will want to put their services on our facilities,” says Stowe.

Beating cable at its own game. Using cable’s technology to preempt cable means moving as fast as cable, which PacBell thinks it can do, based on the low estimated costs of implementing its system. PacBell, because it must replace rather than merely upgrade the existing network, will pay more per unit of construction than cable. But the telephone industry traditionally spends five times as much as cable on construction, leaving more than enough margin to keep pace with cable at PacBell’s projected cost rate.

BELL ATLANTIC DISPUTES THE MERITS OF THE PACBELL PLAN

Ray Smith, chairman and CEO of Bell Atlantic, has much to lose in California if PacBell beats TeleCommunications, Inc., to the punch. But he also has much at stake on the other side of the country, where Bell Atlantic has won strong support among state regulators for the alternative dedicated-star approach to network design. In New Jersey, for example, Bell Atlantic recently filed an accelerated deployment plan that rests on the cost and performance claims it has made for the traditional telco broadband network approach. State support for this expansion includes a considerable amount of revenue-enhancing deregulation, which will allow Bell Atlantic to charge what it wants for services that are deemed competitive at a very low threshold of competitive activity.

A contradiction for Bell Atlantic. Smith fell into an apparent contradiction in a recent attempt to discount the merits of PacBell’s technological choice. Talking to reporters at the Western Show cable convention in Anaheim, CA, earlier this month, he suggested that PacBell was signing off on old technology that would be quickly outdated by the type of network his own company is building in its operating territories.

“A cable system is all PacBell is building,” Smith said. “Those who build analog type networks only to fight off cable companies will be doing the wrong thing.”

But Smith’s remark ignored his commitment to the cable full-service network idea that TCI (the largest cable MSO) embodies — and which Bell Atlantic plans to take over. Indeed, it ignored the fact that TCI technologists, including CEO John Malone, agree with PacBell’s contention that integrating voice, data and video services over a cable-like network is a better way to go for any anticipated applications, extending into the next decade.

WILL CALIFORNIA REGULATORS APPROVE THE APPROACH?

It’s not a certainty that PacBell will actually be allowed to do what it says it wants to do. A lot depends on whether the California Public Utility Commission (PUC) thinks this is the right approach. If it does, other states will fall under tremendous pressure to reconsider their regulatory support of the costlier star/star/star design.

So far, California regulators have not indicated whether they believe PacBell is on the right track. Daniel Fessler, president of the California PUC, says he has doubts about the technical viability of the star/bus approach. But other commissioners are less concerned.

Not another Betamax. “I’m very concerned about having a lot of money invested in California in ‘Betamax’ machines,” Fessler says. “If this architecture fails, a lot of products will fail. It will chill consumers’ acceptance of products if they believe they will be punished through mechanisms of the marketplace for being one of the first individuals to accept the risk of making personal investments in new equipment.”

Fessler says it appeared his fellow commissioners are less concerned about this than he is. But he added that both his and their concerns go further, extending to the issue of “stranded capacity.”

Issues of backward compatibility. “There’s some concern about backward compatibility (between the new network sections and the old),” he says. “There is tremendous investment in, and reliance on, the existing system.”

PacBell executives, however, are confident that once the regulators understand what is going on, the whole scheme will make sense to them. This star/bus network is so superior a choice, they say, that it not only matches the capabilities of any currently available or soon-to-be-available design options, but it does so at costs that can be recouped from savings in operations alone.

SAVINGS IN MAINTENANCE COSTS TO PAY FOR NETWORK

The potential savings from being able to remotely identify fault locations and to take preventive action based on telemetric feedback from any point in the network has long been a driver in the digitization of telephone backbone networks. But the costs of extending digital capabilities to the home have always been considered too great to be justified by savings on maintenance alone. Complete digitization of the network, it was thought, required a significant revenue gain as well.

Greater control with star/bus configuration. According to Lee Camp, PacBell’s vice president of strategic planning, the star/bus system puts a digital, easy-to-monitor signal out in the field all the way to the customer premises at far lower cost than would be possible over twisted-pair copper phone lines or fiber.

“We’ll be able to pay for this network through savings from lower maintenance, improved inventory management and lower provisioning costs,” Camp says.

Growing support for star/bus network. There are signs of growing support for the star/bus argument beyond PacBell. Joe DeMauro, strategic planning director for Nynex, says his company’s rollout of broadband networks (which closely matches PacBell’s in scope and timing) will use the fiber/coaxial hybrid in densely populated areas, while sticking with traditional fiber-to-the-curb in the suburbs.

And US West, after initially opting for the dual narrow/broadband networks described above, is now considering going to an integrated star/bus approach as well, company sources said.

Vendors, too, are flocking to the concept. Leading the stampede is AT&T Network Systems, the supplier for all of PacBell’s broadband distribution gear and a pioneer in development of “cablephone” technology. Other entrants among the telecom giants include Scientific Atlanta, General Instruments Corp. in an alliance with DSC Communications, ADC Telecommunications and Northern Telecom.

AT&T TO SUPPLY PRODUCTS FOR BROADBAND TELEPHONY

The PacBell deal marked AT&T’s first product entry into broadband telephony in the United States, following several years of internal research and debate. According to Jack Harrington, a regional vice president of AT&T Network Systems, the company will provide whatever design approaches customers desire; but at this point, no other broadband product line is slated for the U.S. market.

“We think that, across the globe, customers are going to settle in on some variation of this theme we have introduced here,” says Harrington. Even in Europe, he adds, the present commitment to all-optical systems could change and probably will in some areas.

“We’ll be responsive to what customers want, so we’re not eliminating any possibilities by going in this direction,” he adds. “But we see a tremendous opportunity here.”

PacBell opts for “integrated services approach.” Harrington indicated there are significant distinctions between what AT&T is supplying to PacBell and the type of overlay fiber/coaxial analog network that was derided by Smith. “The type of design you want depends on how you view the migration path to broadband,” Harrington says. “Do you want to build the business around existing services, or do you want to move directly to next-generation fully integrated services? PacBell has chosen the integrated service approach.”

AT&T Network Systems won the PacBell order for everything from the central office-based Host Digital Terminal (HDT) through to the network interfaces mounted outside the customers’ houses. It represents one of the biggest single-vendor contracts ever announced in the telephone industry. The companies said the technology, while similar to the AT&T system developed for the UK cable-TV market in cooperation with Optical Networks International, differs in important respects. It employs proprietary software from Pacific Bell, and it takes an overall design approach that is more compatible with U.S. telephone networks.

Cable operators want to cooperate to compete. Already, according to sources close to California cable operators, PacBell’s plan is stirring cable companies to think about ways to cooperate in network development, where facilities sharing would support service competition. Sources said that with San Jose on the list of early PacBell deployment targets, TeleCommunications, Inc., the holder of the city’s cable franchise, is exploring the possibility of sharing facilities rather than facing the prospects of PacBell completely overbuilding its facility.

“TCI has to spend $70 million on underground conduit alone,” a source said. “PacBell will have to spend even more. Does it make sense to dig separate trenches if you’re pursuing the exact same layout with the same type of wiring?”

This topological compatibility, combined with the high potential capacity of the star/bus system, suggests that regulators could soon come under pressure to permit joint ventures — if not joint ownership of facilities — among competing cable and telephone companies. If the PacBell plan takes hold, a lot of assumptions underlying government resistance to such approaches could fall apart.

Fred Dawson

A LOOK AT THE STAR/BUS NETWORK

Telephone and cable network engineers have different parameters to work with when designing a star/bus system, but the basic elements of the design are applicable to both industries’ existing network infrastructures.

The key facilitating piece of equipment is the HDT (Host Digital Terminal), which is located at the central office in the telco’s case or at the headend or at major distribution hubs in a cable system. The HDT serves as a concentrator and manager of traffic. In the case of both cable and telephone networks, fiber carries the signal from HDT to nodes, which are points of interface between the fiber and coaxial distribution cable that passes all the customers in a service area.

The ordinary telephone network (a star topology) dedicates a separate wire to each customer. In contrast, bus techniques put many customers on a single wire. The customers’ calls are kept separate either by assigning specific time slices to each in turn (time-division multiplexing) or by assigning a separate radio-frequency channel to each (frequency-division multiplexing). Currently, frequency multiplexing appears to offer advantages, such as the flexibility to choose a multiplexing technique that is best suited to each application.

Virtual channels. The design supports dynamic allocation of bandwidth on a “virtual channel” basis, although vendor products vary significantly with respect to these and other details. With fiber penetration to 500-home service areas, all designs claim to offer ample bandwidth for high-contention use of the facilities.

They are also intended to preserve the look and feel of existing services at the customer premises. As Ron Foster, vice president and general manager of telecom systems at Scientific Atlanta, says, “The (voice-over-cable) system was designed from scratch to mimic everything that a telephone system does with no changes in customer equipment.”

The system terminates the coaxial drop at a residential interface unit (RIU) mounted on the outside of the customer’s premises. The RIU directs voice signals to standard twisted-pair lines inside the premises and passes the cable service to the in-home coaxial wiring. The system has no impact on which types of cable security and settop devices are used by the cable operator.

The General Instrument method. In a contrasting approach, Andy Deveraux, General Instrument vice president of strategic planning, says the integrated-service system GI has developed with DSC provides a single unswitched path to the settop from the headend. The customer terminal electronics are located in the settop box instead of the RIU.

The DSC system is designed to meet Bellcore’s broadband-switching standards, which means the box in the home will have access to every type of advanced digital service envisioned for the telephone industry. According to Deveraux, the firms will also offer other versions of the system that establish different locations for the customer interface units.

THE WAY BACK

It is also important to provide transport for return signals from customers over the coaxial-bus portion of the network back to the interface with the fiber trunkline. Some vendors are sticking with the established cable-industry return path, which uses the 5–30 MHz (megahertz) band. Others, like AT&T Network Systems, are moving to wider bandwidth, in the range of 5–42 MHz; this is made possible by a big performance leap in the filters that separate upstream from downstream signals.

PacBell and other users contend that this amount of return bandwidth is more than enough to handle any reasonable level of usage of phone service within the 500-home coaxial serving area. Northern Telecom, in fact, is working on a new cablephone system that it claims enhances bandwidth efficiency to the point that much larger coaxial serving areas, covering as many as 2,000 households, could be provided phone service without deeper extension of fiber.

Fred Dawson