September 5, 1999 

Within a decade, most people in developed countries will have access to Internet connections that are tens if not hundreds of times faster than the ones in common use today. Although that development may not sound exactly earth-shaking, it will in fact herald an entirely new stage in the evolution of that global network.

Those high-speed connections to the home--whether they take the physical form of a telephone wire, a cable television line or a satellite link--will give rise to an entirely new set of applications. Not only will enthusiasts be able to jump instantaneously from page to page on the World Wide Web--and will therefore use the Web much more often--they will also be able to enjoy applications that exist today only as crude prototypes or as concepts in the minds of visionaries and entrepreneurs. Real-time high-fidelity music, telephone, videoconferencing, television and radio programs could all be provided by a single service company over a single hookup. There will be new entertainment options, such as movies-on-demand, and new features, such as the ability to call up information about a movie's director or its actors as they appear on screen. Users will be able to play on-line games--live--against many contestants scattered around the globe. People separated by thousands of kilometers will be able to share virtual-reality experiences and work effectively together on a business or academic project.

The rapid rise of very popular Internet applications like the Web is driving industry to build the infrastructure needed to bring high bandwidth, or "broadband," communications to the home. It is estimated that in the U.S., one home in four now has some kind of access to the Internet. Most people now connect using a dial-up modem, a device that converts back and forth between streams of data and patterns of audible-frequency tones, enabling the data to be sent down telephone lines originally designed to carry voices. At the other end of the line, an Internet service provider acts as a kind of portal through which the subscriber can contact and exchange data with countless nodes around the globe.

Dial-up modems are easy to use, and most computers come with one built in. But their performance is limited. In addition, the need to place--and pay for--a telephone call to establish a connection to the service provider means that access to the Internet is not continuously and conveniently available. Once the broadband technologies described in this special report emerge, people will receive and send text, images, audio and video over the Internet at vastly greater speeds, with virtually no delays. Indeed, the Internet will always be "on" at various screens in the home, ready to help at the single click of a key or voice command. The technologies will bring a host of new data, multimedia and television services. And they will do all this at an affordable price.

Wiring the Home 

The difference in speed among ways to connect to telecommunications services such as the Internet is striking. The fastest modems in general use today receive and transmit data at 56,000 bits per second (56 kilobits per second, or kbps). This is the limit at which most home computers can connect. The personal computer in an office is typically connected to others in the company with a local-area network; the most common, Ethernet, has a raw speed of 10 million bits per second (10 megabits per second, or Mbps)--about 200 times faster than the modem. But unless the company has a dedicated high-speed line to an Internet service provider (ISP), the worker's Internet experience is limited by the modem, too.

Some companies have high-speed connections, but very few homes or small businesses do. Furthermore, people do not usually leave their computer and modem on all day, connected to their ISP all the time; the charges for this continual use of a phone connection would be prohibitive. This non-continuous connection has two implications. First, when people want to use the Internet, they must wait while the modem connects, which hinders casual or frequent use. Second, there is no way to exploit applications such as receiving a phone call over the Internet, because the recipient cannot be contacted if he or she is not already connected.

Ultimately, the desire for broadband communications to the home derives from the increasing speed of computers. Computers are on a development path that improves performance by a factor of 10 every five years. This steady advancement regularly prompts a wide range of unanticipated applications; the World Wide Web was basically just a bright idea only nine years ago. Yet the increasing speed of computers will not result in faster communications applications if users remain stuck behind a dial-up modem; 56 kbps is about as fast as this hardware will ever go.

The assortment of wires and cables that run to most homes also represents a barrier to high-speed operation. None of these links was intended for data transmission at any speeds, let alone extremely fast ones. Twisted pairs of copper wires were put in to support phone service; coaxial cable carries television signals; and power lines, of course, convey electricity.

Nevertheless, engineers are trying a variety of approaches to connecting homes for high-speed data communications; this report describes five of them. The first two use clever technologies to wring the most out of existing wires to the home: hybrid fiber-coax makes use of the cable TV industry's infrastructure, which includes fiber-optic lines in addition to coaxial cable; digital subscriber line, meanwhile, exploits frequencies much higher than those used to convey conversations to send high-speed data over pairs of copper telephone wires. The third approach is to run an entirely new wire to the home--fiber-optic cable. There are several configurations for such a system, including fiber-to-the-curb and fiber-to-the-home, depending on how far the fiber reaches toward the residence.

Wires and fibers can be abandoned altogether. The fourth and fifth technologies are both "wireless"; in addition, they each have an analogue in the wireless telephony arena. Various planned broadband Internet-oriented satellite networks would work in a manner similar to the Iridium satellite-telephone system in that the orbiters would communicate directly with the subscriber. In the case of the Internet system, the user would access the data via a small dish antenna. The other approach, known as local multipoint distribution services (in Canada as local multipoint communications system), is similar to a cellular telephony network; it uses radio waves to transmit data between towers and receiver dishes mounted on homes.

There is no single, simple metric to compare these broadband technologies. It would be nice to rank them on comparative speed, for example, but almost all of them are capable of operating over a range of speeds, depending on how they are actually implemented.

By considering the key features of each, we can at least make reasonable inferences about likely answers to some of the fundamental questions, such as: Will one technology win out over the others? Will several compete? Will broadband service be more affordable anytime soon? Extremely powerful industry forces now at work will settle these issues.

Next on Channel 92: Lightning-Fast Internet 

Although the cable TV industry developed its widespread coaxial cable network just to offer television, it aggressively upgraded its equipment starting in the late 1980s to support other services, including Internet access and even telephone service. The enhancements involved laying fiber-optic lines from key signal distribution points most of the way to residential areas, then using the original coaxial cable to distribute the signal among the homes in a neighborhood or part of a town. By using fiber-optics only where it was most needed, the cable companies spent far less than would have been necessary to replace the entire network with the optical lines. Nevertheless, even this partial use of optical fiber improved the television signal while making it possible for the network to carry two-way Internet and telephone traffic. To access the Internet, the homeowner must have a cable modem, a device that attaches to the cable just like a TV converter box but decodes and manipulates data rather than television signals.

The capacity of a hybrid fiber-coax (HFC) system is considerable. Just one of the many television channels offered to subscribers can carry almost 30 Mbps to the home. Moreover, technicians could allocate multiple channels for broadband Internet if the demand generated enough revenue to justify displacing other television channels.

In an HFC system the data channel is shared among the homes linked by coax to the end of the local fiber-optic line. Thus, the actual data rate achieved in any individual home depends on the number of users sharing the channel at a given time. But a well-designed system can give each user bursts of data from the Internet at speeds around 10 Mbps. There is also a lower-speed channel in the reverse direction to carry data from the home back to the Internet.

From the Telephone Switch to You 

The telephone industry has developed a number of novel techniques to transmit data at high rates over the copper-wire pair designed to convey phone calls. For example, a service called integrated services digital network (ISDN) has been around for years; it operates at 64 or 128 kbps. But a pricing quagmire and an introduction long confounded by regulatory issues left it only marginally more useful than fast modems today. A much faster service known as T1 was initially developed for bringing multiple voice connections to a business; it can carry data at 1.544 Mbps, and some small businesses and even home offices have begun using it for data access. Still, T1 has traditionally been priced for commercial voice access, which is much more costly and more than most people can afford for data access.

The current promising telephone technology is digital subscriber line (DSL), which operates over conventional phone lines but achieves higher data rates using different electronics at the ends of the wire. The twisted pair from a home typically runs to a building not far away called a central office, where it connects to a switch. A switch is a complex piece of equipment that routes telephone calls to other switches or phones as necessary. Most of them were designed only to carry voice and have no special options to handle high-speed data. Dial-up modems work only because their designers went to great trouble to create coding schemes for data that the existing switches can carry.

DSL is much faster because it does not use the existing switching equipment. New switches are installed in the central office to exploit the full data-carrying capacity of the wires, which normal phone calls simply cannot use. DSL also uses more sophisticated schemes that code the data in a bandwidth that is larger and occupies much higher frequencies than the one used for voice.

There are several variations of DSL, depending on the distance from home to central office. At present a home must be within about five kilometers of the central office to make use of this scheme. The most widely developed version is asymmetric DSL, or ADSL. It is capable of delivering 3 to 4 Mbps to the home and a slower rate back from the home, typically a small fraction of a megabit per second.

Fiber-optic lines have a number of advantages over copper pairs or coaxial cables. Most important, fiber can carry data at a much higher rate: millions of megabits per second. If optimized, a single fiber could carry all the phone calls being made at any instant in the U.S. Today there are hundreds, if not thousands, of fibers nationwide that serve as the backbone of existing telephone, cable TV and Internet networks.

 

HIGHER-SPEED ACCESS...makes the Internet feel more responsive by reducing the time for data transfers. The curve shows how access speed affects the total time needed to download a complex Web page, with included images and subcomponents, from a site across the country. Obtained by a detailed simulation of network operation, the curve shows that for this example a 56-kbps modem might take more than 40 seconds, whereas the broadband technologies need just under 10 seconds. Not all the broadband options are shown: for most PCs, the 10-Mbps Ethernet link sets the speed limit, and even fiber-to-the-home (off the scale at 100 Mbps) cannot beat about six seconds set by network processing times and the speed of light.

Signals are sent by shining a beam of laser light down a fiber. Because of the optical qualities of the fiber, the light can follow the twists and turns of the strand, coming out the other end where it can be detected. The laser is turned on and off at a rate of billions of times a second, generating a pattern of light pulses that correspond to bits and are sent down the fiber and converted at the receiving end back into an electrical signal.

As with any kind of infrastructure, it is very expensive to install a network linking many homes. One way to reduce the cost is to use one fiber to serve a cluster of residences, rather than having a separate fiber for each home. This cheaper system, which seems to provide a fairly good cost/performance trade-off, is called fiber-to-the-curb. One fiber comes from a central office to a small box near the curb, and from there the traditional copper pairs or coaxial cables connect to perhaps 10 or 15 homes.

Will LEO Roar? 

Of all the broadband options now emerging, satellite-based service is the most advanced and the most risky--from both a technical and an investment perspective.

Most existing communications satellites are geosynchronous. This means they are at exactly the right height above the earth so that they orbit at exactly the same rate as the earth turns. The result is that they are fixed in the sky with respect to a receiver on the earth. Thus, home satellite dishes for receiving TV signals don't have to move to track the satellite.

Yet geosynchronous satellites have several drawbacks. They are a long way up (about 36,000 kilometers), so there is about a quarter-second delay in sending a signal up and back. This delay degrades many forms of data transmission. The distance also means that the satellite must have a powerful transmitter or transmit at a low data rate. Finally, there is only so much room in geosynchronous orbit, and it is already mostly full.

The next generation of satellites being readied for deployment will have much lower orbits. Instead of seeming stationary, they will pass by overhead. If enough of them are placed in orbit, at least one will be in range at any time over any given point. These low-earth-orbit (LEO) satellites will also be engineered to communicate with one another [see "New Satellites for Personal Communication," Scientific American, April 1998]. This way, a remote unit at one residence can talk to another unit somewhere else by sending data up to whichever satellite happens to be overhead at the moment and having that satellite forward the message around in space to the satellite that happens to be over the second remote unit.

LEO systems have many potential advantages. Because the orbital altitude is typically below 2,000 kilometers, the propagation delays are very short. Because the satellites can operate at various altitudes, more orbits are possible, and many systems could be deployed. The low orbit also means lower radio transmission power, so the home needs only a small, unobtrusive antenna.

The cost of putting a LEO satellite system in place is very high, because dozens of satellites have to be launched. And it has yet to be demonstrated that demand justifies this cost. Nevertheless, LEO voice systems such as Motorola's Iridium configuration are already in place, and several companies are planning LEO data systems with projected aggregate data rates up to one gigabit per second.

Wireless on Terra Firma 

Broadband service can also be ground-based and wireless, and indeed those are the distinguishing characteristics of a diverse set of technologies being explored for the consumer marketplace. This report focuses on a particular option, called local multipoint distribution services (LMDS), which is receiving a lot of attention from access providers. LMDS systems use a radio signal of very high frequency (28 gigahertz).

The basic premise behind wireless networks is that the major cost of installing any broadband system based on wire or fiber is not the cable itself but the labor to install it. Thus, they eschew wire lines. Instead, like cellular telephones, these networks use radio connections from a base station antenna to remote units at residences.

Engineers are developing a number of configurations, which can be categorized by the distance between base stations, the data transmission rate and issues such as whether the remote units can be mobile. Unlike cellular telephones, Internet users are generally stationary, which greatly simplifies the system.

In fact, one such wireless system for data is designed to use the existing cellular telephone towers and thus must operate using the spacing of those facilities. So far these systems offer only modest data rates (10 to 50 kbps) and are marketed to users on the move, not for residential access. Still, the cellular industry has plans for more aggressive use of its towers, to deploy services with data rates up to 1 Mbps to the home. Commercial service is expected in a few years.

Other wireless systems are based on much closer spacing of base stations. Smaller antennas would be put on top of telephone poles or even hung from lines between poles. These systems would be more costly to install, because they would require more base stations but could offer higher data rates given that the wireless distances would be shorter.

The very high frequency of LMDS imposes some limitations, because the radio waves travel only in straight lines and are therefore blocked by buildings and other obstacles. Even more problematically, they cannot penetrate moisture and thus cannot go through foliage very well. But the very large allocation of bandwidth (1.3 GHz) allows for the potential creation of very high speed services to the home. Early trials of LMDS systems are happening now, involving perhaps 10,000 subscribers in the U.S.

How Fast Is Fast Enough? 

It is clear from these brief descriptions that the various broadband systems differ in the performance and range of services they can provide. For example, ADSL does not have the capacity to carry television, so it is suitable only for data and voice. Some forms of fiber-to-the-curb support TV service, some only voice. The various satellite systems currently being deployed are specialized for voice, data or television.

These systems have been in the works for a while, so the fast rise of the Internet has imposed a new set of design challenges. A telephone call requires a known quantity of network capacity, and a system to carry telephone calls can be designed for an expected number of calls. Similarly, a cable television system is designed to carry a known number of television channels. But the speed of Internet service can vary greatly. So right now system designers and broadband entrepreneurs are left to figure out how much money people will pay to go faster--which basically boils down to determining how much happier higher speeds make them.

Designers are also plagued by uncertainties in future user demands. The most popular application of the Internet today is the Web, which has one set of capacity requirements. But the Internet can carry all kinds of applications, each with different communications requirements, and it is not clear which will become popular over the life of a system installed today.

Given the wide range of systems and technical requirements, it is tempting to speculate which broadband technology will be the winner. But technical differences will have minimal influence on deployment. The various systems are all technically feasible. All of them, except LEO satellites for data only, have been demonstrated and are even being installed to varying degrees. The real barrier to widespread broadband access is the cost of installation. As might be expected, economics and business structure are driving deployment decisions.

Industry estimates suggest that the cost to newly wire a neighborhood for broadband, so that the cost of installation is shared across all the residences, is roughly $1,000 a home. Because there are about 100 million homes in the U.S., the implication is that perhaps $100 billion must be spent to provide a new wire-line connection to every home in the country. This is a huge sum to justify, especially because the importance of the Internet (and broadband access to it) is still proving itself. To wire one home at a time is much more costly, so any piecemeal wiring of isolated consumers would be even less feasible.

Telephone and cable companies are therefore moving forward with incremental improvements that are substantially cheaper than a total replacement. Today, not surprisingly, cable television companies are selling broadband Internet service based on their hybrid fiber-coax networks and cable modems. Meanwhile telephone companies are offering broadband Internet service based on ADSL and their ubiquitous copper pairs. The eventual mix of cable and DSL technology in the U.S. will thus have nothing to do with their relative technical merits but everything to do with the relative levels of investment and marketing by these two market forces.

There are few investors who will back the installation of a whole new wire-line facility such as fiber-to-the-home. But when a new installation is called for, say, to serve a new subdivision, there is considerable motivation to install as modern a system as possible. Thus, fiber-to-the-curb may be seen in new installations, most likely brought by either the local phone or cable company. Intriguingly, fiber-to-the-curb or fiber-to-the-home may also enable electric utilities to enter the data business, at a time when many are seeing successful challenges to their former monopolies on power generation. One tremendously valuable resource many electric utilities still have are rights-of-way--which are, basically, the poles and lines that run up and down neighborhood streets and major thoroughfares. It would be a fairly straightforward matter to string new data communications lines alongside the old power cables. The possibility underscores the idea that each of the wire-line technologies should be seen as occupying a business niche, rather than competing with one another on technical merits.

Companies that want to get into the broadband business but that do not already own a wire to the home or right-of-way are left with wireless or satellite. Erecting these systems, particularly in municipal regions with high population densities, is less expensive than putting in new wires but still very costly. Consider AT&T, which as a long-distance company is not wired directly to residences. AT&T has committed more than $100 billion in acquiring cable companies such as giant Tele-Communications, Inc., to give it broadband paths to some portion of American homes (ironically, via TCI's extensive cable television networks). To serve other parts of the country where these mergers do not give it wire-line access, AT&T is developing wireless options.

Clash of the Titans: Cable versus Telcos 

Broadband technologies have not appeared in the marketplace as rapidly as some observers had hoped, and frustration with slow deployment has led to speculation that perhaps there is not enough real demand to justify the large investments required. The Internet's quick rise will speed things up. Deregulation of telecommunications was supposed to have encouraged phone companies to offer television and cable companies to offer phone service, thereby improving competition in each sector. Little has actually happened. But huge demand for the Internet, which can be carried just as readily by both industries, is prompting them to collide for the first time.

Whichever industry invests the most in infrastructure might very well grab the lion's share of the broadband market. At the moment, the cable companies, with their hybrid fiber-coax technology, have the largest share. But the telephone companies have the capital--and now, with DSL, the technology--to make an impression in the market and are finally starting to do so. It is too soon to say whether wireless and satellites will mount a significant challenge to cable and DSL, and fiber-to-the-home seems destined to remain a prohibitively expensive proposition for at least the near term.

In the longer run, the consumer will be best served if all of them succeed, fostering more vigorous competition and more extensive choice. It is one of the Internet's greatest strengths, in fact, that it can operate over all these technologies. Happily, there does not need to be one winner to take us to the next stage in the Internet's evolution, regardless of the specific shape that stage finally takes.