The early vision of fiber-to-the-home, which promised consumers abundant, ubiquitous and future-proof bandwidth, has remained largely unrealized. The basis for the original vision of FTTH has been overtaken by technological, regulatory and economic shifts.

We believe the realities of today make it unlikely that the major players will build out a network without first seeing demonstrable demand for it. The current climate suggests that only "green-field" developments can be sustained. On the other hand, service and societal changes could accelerate the process, government intervention or subsidization could motivate investment, and business strategies-coming from players without a telecom embedded base to protect-could emerge that would realize the promise that FTTH holds.

Essentially, the vision of fiber-to-the-home was developed to satisfy the perceived need for future consumer applications. Optical fiber, it was argued, would permit high-bandwidth transport, remove bottlenecks as video-rich services were developed, enable upgrades and permit passive multiplexing that would remove remote-powering costs.

In today's terms, applications such as telephony and video-on-demand were to be carried over a B-ISDN universal carrier at OC-3 rates. Proposals included an optical version of the telephony loop (each fiber carrying one house's circuit) to passive optical networks (PONs) with an optical splitter delivering light (and bandwidth) to several optical-network units (ONUs). For fiber-to-the-curb (FTTC), the ONUs served several homes.

In the nearly two decades since the FTTH/C vision was established, however, economic, regulatory and technological forces have frustrated the vision's intended universal implementation.

A number of structural forces have been at work. First, the two major consumer applications-broadcast television and telephony-have nearly opposite characteristics. TV is a largely passive, impersonal entertainment delivered as a one-way service requiring high bandwidth. Telephony is personal communication that requires a two-way symmetric, low-bandwidth connection. As a society, we apparently value communication more than "content" because the cost we pay per bit of information on phone calls exceeds that of entertainment by six or seven orders of magnitude.

TV's infrastructure is largely coaxial cable, with multisystem operators (MSOs) as service providers, while telephony is largely delivered over twisted wire pairs by incumbent local-exchange carriers, or ILECs. The two-way nature and quality of telephony make aggregation, multiplexing and switching essential, and thus the telephone and TV physical networks differ in marked ways.

Both MSOs and the ILECs have, over time, created efficient networks, and the providers can be viewed as vertically integrated: They own the networks that deliver their services. Both entities saw potential markets in the intermediate area of data communications, which could provide personalized entertainment, be consumed as an end in itself or enhance communications services. This seemed a natural extension for the ILECs, since those services required point-to-point connections, while regulatory and technical reasons prevented MSOs from entering.

This space was intended for FTTH: The services would need more bandwidth than the copper plant allowed and would need a new architecture, but revenues from new services such as video-on-demand (VOD) were expected to support a fiber buildout. Developments in the 1980s, however, dramatically altered the landscape. Advances in linear lightwave technology enabled CATV trunks with cascaded amplifiers to be replaced with high-bandwidth, high-fidelity optical analog transmission links to local nodes.

The "pushing fiber deeper" approach became known as hybrid fiber/coax (HFC), which, with its ability to shrink serving-area sizes, began to make it appear that the telcos' FTTH/C plans would permit the MSOs to carry data (albeit on a shared medium). This would crowd into the telcos' vision and could lead to competition in the new segment. Meanwhile, technical advances in data compression reduced the information needed to carry video by nearly two orders of magnitude, and new modems with a variety of digital subscriber line (xDSL) formats were developed to carry data and video over copper pairs.

Finally, user-friendly graphical interfaces made Web traffic explode, launching the dot-com boom and competitive local-exchange carrier (CLEC) threats. Technical advances also made satellite TV a credible threat to cable, and wireless phones a threat to land lines.

Thus, technical progress and regulatory relaxations set the stage for that middle area to be actively contested, while putting pressure on each market.

So, what does the future hold?

Overriding all of the plans, of course, is the need for the network to make money for its owner: Early advocates estimated that the customer's willingness to pay for phone and VOD services would support a cost of $1,500 per subscriber for infrastructure, which is still a reasonable guess today. Activity in forums and standards groups such as the full-service access network (FSAN) and Ethernet-in-the-first-mile (EFM) activities support the vision, and vendors are within striking range of that cost. Furthermore, several large trials, dating back to the 1980s, have already been run by ILECs, and some MSO trials have recently begun, so the players are developing the requisite technological experience.

What are the prospects for a massive FTTH rollout? We believe that a major deployment is unlikely for the foreseeable future, for several reasons.

First, the cost calculations require essentially monopoly conditions. The major cost of the FTTH network is in the installation of ubiquitous distribution fiber that permits a drop to any potential customer. That fixed cost of passing every customer must be borne by the fraction of customers who actually subscribe. If the "take rate" in a competitive market is 25 percent, for example, then the effective cost per customer is nearly four times the cost in a monopoly situation.

Costs for today's services are well-known, and markets are competitive for new entrants. Thus, new providers would have no pricing power-and lenders or financial markets will not permit the stranding of capital in the new, post-telecom-crash environment.

That favors greenfield approaches in which the FTTH build can be justified on costs and the fiber's capacity would dissuade a competitor from trying to overbuild. A second network would presumably cap the take rate at 50 percent for each. Thus, tied to low capital turnover rates, it would take 30 to 50 years for FTTH to become prevalent. Even then, a monopolist owner would have to subsume its competitor's business-a challenge for businesses as disparate in their technologies and practices as entertainment and telecommunications are.

No killer app, yet

Second, the calculations assume the existence of a killer app in video and/or data. HDTV, a slow starter so far, is unlikely to fill the bill: Spectral efficiency and node splitting will permit MSOs to carry it over HFC networks (although it may motivate ILECs ), especially if the HDTV is primarily used only for new movies. Data applications in the form of, say, 80-kbyte Web pages are unlikely to strain either network, because of statistical multiplexing, and recent communications trends (messaging) place more value on mobility than bandwidth. Thus, no application is visible that demands FTTH.

In addition, the capital constraints on new "builds" will likely be tight. It is unlikely that large amounts of money will be sunk into infrastructure that has an unproven killer app. That is, it is more likely that carriers will attract capital to defend their turf and incrementally improve their network than to build a network that might be able to serve an unproven service. The Field of Dreams approach ("If you build it, they will come") is unlikely to attract capital.

So, while we have argued that massive FTTH deployment is unlikely in the near future, there are several ways that the situation could change to increase the chances for its deployment.

First, a killer app could emerge. The best example of such an application would be personal video-such as video phones-that require significant, continuous upstream bandwidth. While earlier trials of this application have failed, more people have experienced streaming audio and video clips in recent years, and such familiarity could stimulate the application. In a similar vein, the emergence of "Ethernet appliances" and home networking with 100-Mbit/second interfaces might stimulate a demand for more-symmetrical, higher bandwidth.

Relinquishing power

Second, the government could form an initiative that would support broadband to the home. Such a move would be an economic policy, motivated by the correlation between the communications infrastructure and the GDP. As occurred for electrical apps when the power grid was in place, for instance, broadband applications that have yet to be conceived or that have been considered insufficient to justify an initial buildout could, after that buildout occurred, become potent economic drivers.

Also, an independent third party, eschewing vertical integration and instead separating the network from the infrastructure, might build a network capable of efficient delivery for all services and then offer it at "arm's length" from the services. By relinquishing the hold of an existing service, such a player would enable a whole suite of services to be delivered by other providers.

Each major service player would then be forced either to join the network or to watch other players move into its core business. To avoid stranding capital, such a player would probably have to be a utility that would not need to negotiate rights of way.

The complete version of this article was presented at the 2003 Optical Fiber Communication Conference.

N. Frigo, K. Reichmann and P. Iannone are researchers at AT&T Labs (Middletown, N.J.). Send e-mail to frigo@att.com.

http://www.eet.com