High-speed access assessed: Fiber may be poised to move closer to the home, where it can support numerous high-bandwidth applications

For years, business and residential users have relied on copper wire to deliver voice and data communications to their service providers. Data that Communications Industry Researchers recently compiled, however, indicates that fiber will play an increasingly important role in the access arena. Driving this change will be customers' seemingly insatiable appetite for bandwidth.

The high-speed access environment in North America can be divided into two segments. The low-end segment, providing access from T-1 to T-3 speeds, is sometimes referred to as "midband" to distinguish it from the true broadband segment, which lies above the T-3 level and is occupied by mostly experimental technologies.

Essentially, two types of technology provide midband access. Legacy local area networks and 25 Mb/s asynchronous transfer mode provide midband access at the desktop, while the emerging digital subscriber line (DSL) and hybrid fiber/coax (HFC) technologies provide this level of access to both the desktop and premises concentrators. Satellite and wireless services such as local multipoint distribution service (LMDS) may soon challenge DSL and cable modems.

Initially, the industry predicted that interactive TV would drive the demand for high-speed access technologies. This service has had disappointing results, however, and the same DSL and HFC technologies that originally were intended to carry video into the home have been refocused as a means of connecting to the World Wide Web.

There is also an increasing tendency to focus on business rather than residential applications. In the many interviews that CIR conducted, vendors pointed out that businesses, not consumers, have the money to spend on new technology and that it is typically easier to justify such expenses in a business to support applications such as telecommuting and Internet access.

Broadband access encompasses a number of high-speed LAN technologies. But for the time being, access to wide area networks at speeds above T-3 is confined to the relatively few large corporations that buy Sonet OC-3 (155 Mb/s) or ATM access to the Internet or a private network.

DSL or HFC? DSL is the high-speed access technology that has attracted the most attention among telcos. It comes in a variety of flavors (Table 1) and promises relatively high-speed access over existing copper loops.

Furthermore, DSL technology appears to provide point-to-point access to the Internet that does not tie up the telco switch the way that dial-up POTS and ISDN access do. While most of the current DSL versions operate at a few megabits per second, the technology promises a migration path to systems that operate at 50 Mb/s.

Not surprisingly, the Bell companies have been extremely enthusiastic about DSL technology. Some of them have formed a consortium to provide a standard architecture for asymmetrical DSL (ADSL) deployment, which is the most hyped version of the technology. This consortium has initially settled on Alcatel's version of ADSL.

In the meantime, the Bell companies have run trials and are now beginning to make commercial deployment announcements.

U S West's !nterprise division became the first major U.S. data service provider to offer commercial DSL services when it recently introduced relatively low-speed services (up to 704 kb/s) in 14 cities. These services are offered as end-to-end managed services that include all the necessary customer premises equipment.

Meanwhile, Ameritech says it will make ADSL widely available in 1998. Bell Atlantic plans to begin offering an ADSL-based data service for consumers in mid-1998, followed shortly thereafter by a business offering. Before its merger with Bell Atlantic, Nynex said it would deploy commercial ADSL services gradually throughout its service territory based on customer demand.

SBC Communications has begun a limited rollout of ADSL services, and its merger partner, Pacific Bell, plans to begin offering two ADSL services in early 1998-one aimed at consumers and small businesses, the other at telecommuters. Pacific Bell also is beginning to offer new DS-3 and ADSL access options for its frame relay and ATM services.

GTE is considering both 1.5 Mb/s and 4 Mb/s ADSL services.

Interest in ADSL is not confined to local exchange carriers. With DSL now refocused on the Internet, many Internet service providers also are beginning to take notice. WorldCom's UUNet subsidiary is offering a service based on ISDN DSL (IDSL). UUNet's Preferred Access service is being rolled out across the U.S., beginning in California. It gives business customers a 128 kb/s dedicated connection to the Internet. WorldCom's competitive access provider subsidiary is also offering a similar IDSL-based service.

Smaller ISPs that are beginning to offer DSL-based services include Boston-based HarvardNet, which is offering a symmetrical DSL service at speeds up to 768 kb/s, and InterAccess, which offers ADSL at 1.5 Mb/s downstream and 64 kb/s upstream in Chicago.

NetSpeed, which began shipping its ADSL modems only in January 1997, has lined up more than 60 ISPs to provide high-speed access to the Internet.

Meanwhile, Paradyne has agreed to provide DSL equipment, including its HotWire multiplexer, to Network Access Solutions, a CAP in the mid-Atlantic region. The carrier will use the DSL gear to connect its network to ISP networks for high-speed data connections to PCs.

With DSL technologies finally beginning to generate revenues for service providers, it would be logical to predict a future in which these technologies are dominant. Some Bell companies certainly see it this way. Bell Atlantic, for example, has said that it expects ADSL service to become the leading high-speed network access service in the consumer market.

Perhaps as a result of DSL, several Bell companies have scaled back ambitious plans to deploy alternative high-speed architectures based on HFC, an enhanced cable TV network architecture.

For example, several years back, Ameritech announced a grand plan to provide broadband service throughout its region using HFC architecture. This ambitious plan would have cost Ameritech approximately $400 million, with almost 40% of this investment being deployed in the Chicago area. However, while still active in obtaining conventional cable TV franchises and offering conventional cable TV services in several metropolitan areas, Ameritech now seems to have scaled back the range of services it planned to offer.

Pacific Bell once committed itself to spending $16 billion to rebuild its networks using HFC architectures in its California First program. It now looks like Pacific Bell will abandon most of its plans in the wake of its merger with SBC.

The natural constituency for HFC, of course, is not the Bell companies but rather the cable television companies. For a year or so, the trade press was hyping the story that cable companies, blessed with an installed broadband architecture, would soon bury telcos by offering high-speed access over cable modems. When some of the cable industry's more grandiose plans failed to materialize, the rumors started to fly that the cable companies just did not have enough money to roll out such services.

The truth seems to be somewhere in between. Most of the major multiple systems operators now are beginning to slowly and quietly deploy commercial Internet services based on cable modems across the U.S. It is important to remember that while DSL is not widely deployed at the present time, approximately 15% of U.S. homes are served by HFC networks, although for the most part these networks are still one-way systems.

Cable companies also look at the high-speed access business from their strong position as a content provider and packager. This orientation is particularly obvious in projects such as @Home and RoadRunner, which are designed to roll out a broadband infrastructure for broadband content that the cable industry will develop and sell in conjunction with specialist content providers.

The telephone industry also has made a few moves in this direction, but it has not been particularly successful. The Bell companies have not entirely ruled out cable modem deployment, however. BellSouth, while making some level of commitment to using ADSL for enhancing its existing infrastructure, also has two HFC trials underway.

DSL solutions are a great leap forward, but they represent an interim technology. Future technologies will need to be considerably more fiber-rich to support the kind of bandwidth that the multimedia-enhanced future will require (Table 2). Put another way, the future will look more like HFC than DSL.

Interestingly, the most advanced incarnation of DSL-very high bit-rate DSL-also involves bringing fiber closer to the premises, and it essentially merges some of the modulation research that gave birth to DSL with the architectural vision of HFC.

The shift to fiber already has begun, almost as quietly as-but much more slowly than-the commercial rollout of cable modems. Bell Atlantic, for example, has selected Lucent Technologies and BroadBand Technologies to supply the key components of a switched broadband network, also known as fiber to the curb, for locations in the mid-Atlantic region. Bell Atlantic could invest several hundred million dollars in these networks, which are expected to pass 2 million homes.

BBT also is partnering with Lucent to deliver broadband services to 47,000 SBC customers, although this trial may end up a victim of the SBC/Pacific Bell merger.

Broadband without wires Having had their fingers burned with fiber-to-the-home systems in the 1980s, broadband investors now are looking at wireless broadband solutions. These seem to offer a relatively low-cost and quick-to-install infrastructure. Among the solutions currently being offered are interactive digital versions of the old cable wireless systems and especially the newly licensed LMDS, which is specifically intended to be a broadband solution.

BellSouth has run a six-month technology trial with Texas Instruments using LMDS to deliver digital video services to customers in a suburb of Atlanta. In this trial, which also involved an Atlanta-based real estate firm, 25 apartments in the Dunwoody area received 32 video-on-demand (VOD) channels and up to 160 broadcast and near-VOD channels.

Meanwhile, CAI Wireless Systems has tested its multichannel multipoint distribution service and has created a wireless laboratory in Pittsburgh to test two-way voice, video and data services.

Satellites also should not be ruled out as a way of bringing high-speed access to the premises. After all, today's direct broadcast satellites effectively pump 6 Mb/s of digital video into the home.

Typically, satellites provide one-way broadcasting rather than interactive access to the Web like that provided by cable modems and DSL. However, Hughes Electronics, one of the companies behind DirecTV, already has devised a satellite-based Internet access approach for small business and home installation. Called DirecPC, this system offers downstream rates of 400 kb/s, with burst rates up to 12 Mb/s.

DirecPC uses a telephone hookup to send requests to the Internet, then uses the satellite to send the signals back. While this service seemed to be aimed initially at the residential market, Hughes, together with Cisco and Helius, also has announced the availability of DirecPC for NetWare, a server-based information delivery service for many applications, including shared Internet access and file distribution.

The next stage in developing Hughes' Internet access appears to be Spaceway, a satellite that will offer 92 Mb/s of shared Internet access directly to homes via geostationary satellites. Spaceway's two-way digital dishes will cost less than $1000 and could be in service by as early as 1999.

Another planned broadband satellite venture is planned by Alcatel Espace, which has filed with the Federal Communications Commission for a permit to launch and operate a 64-satellite, low earth orbit constellation named SkyBridge (formerly known as Sativod). Skybridge is expected to be operational in 2001 and will provide high-speed broadband interactive services to business and residential users around the world. In addition to Internet access, the SkyBridge system will provide bandwidth on demand for other types of high-speed data communications at speeds up to 60 Mb/s.

While broadband without wires is very appealing, it may have some limitations. Terrestrial solutions, in particular, must demonstrate a very high level of spectrum reuse to provide anything like the bandwidth that can be offered by fiber-rich technologies.

Satellite-based solutions will have to offer the low-cost two-way dishes promised by Spaceway. These capabilities may well be available in a few years, but high-speed copper and fiber access solutions are further along the development and deployment curve than wireless and satellite alternatives.

The large business market For larger businesses, the high-speed access issue has less to do with DSL, cable modems, wireless and satellites than with the future of private lines. Although ADSL provides data rates that are about 30% faster than T-1 at about one-third the cost, ADSL users, unlike T-1 users, will have to be within 18,000 feet of the central office. Meanwhile, virtual circuit technologies are now at a point where they offer a serious commercial alternative to private-line access for many businesses.

Among these virtual circuit technologies, switched multimegabit data service has flopped, but frame relay is hugely successful. And for access at speeds beyond the T-3 level, ATM is a masterful alternative.

ATM offers scalable bandwidths up to almost any level and potentially provides end-to-end connectivity. It also may be able to carry more kinds of traffic than virtually any other kind of networking technology.

Because of the bandwidth demand explosion expected to occur in the next few years, ATM may be the only long-term solution for both access and backbone networks.

A market for private-line service will always exist because some large organizations in the government and private business sectors will constantly want to transfer large amounts of data between two locations.

Sharing bandwidth within the carrier network, as required with virtual circuit technologies, is not acceptable for the very high-end user, especially if usage-based pricing is implemented. Other users may continue to prefer private-line services because of their security.

Even taking such needs into consideration, CIR believes that the stock of private lines will fall to between 20% and 30% of their current levels within 10 years. This will occur because the number of organizations that really need full-time bandwidth between locations are few and far between, and because security measures for virtual circuit networks are now being dealt with by the ATM Forum and other groups. Indeed, strong encryption technology could be applied today to data traveling over ATM and frame relay networks, which would guarantee the integrity of data to a level that would be acceptable to virtually anyone outside the intelligence community.

Clearly, service providers cannot hope to price virtual circuit services at a level that would protect their installed base: If they do not provide ATM and frame relay service at a reasonable price, their competitors surely will do so.

Nevertheless, important new opportunities may arise in the private line market. After all, using private lines to establish full-time connections to the Internet is a market that barely existed until recently.

Additionally, private lines will increase in bandwidth. Multiple T-1s between corporate locations were almost unheard of a decade ago, but they are now mainstream. Private Sonet networks are also beginning to appear and may themselves be mainstream by the end of the next decade.

As high-speed access becomes more of a reality for both business and residential access, new issues will emerge. The nuts and bolts of network deployment is likely to be a primary focus, as will network availability, ease of configuration, installation and interoperability. Users also will demand assurances that promised data rates are actually achieved.

In the long run, these issues-which are precisely the ones that have plagued ISDN for several years-may be more important than price. Some early ISDN users went through a lot of trial and error with service providers before they got what they wanted and will be wary when moving to the next generation of access technology. Nevertheless, high-speed access should generate significant revenues in the U.S. over the coming decade.

Initially, most of the interest in high-speed data-especially at the very highest data rates-will be in the business market (Figure 1). During the next 10 years, however, residential and small office/home office (SOHO) applications will represent an increasingly large portion of high-speed access lines, as fiber reaches closer and closer to residential and SOHO users. A summary of CIR's high-speed access report is posted on the company's Web page at www.cir-inc.com.