BY BROUGH TURNER

While the Internet backbone has been growing by leaps and bounds, there's a real bottleneck in individual access. Data connections for home users remain far too slow.

It has been technically possible to make a data connection from the home for decades. In 1968, the Bell System first offered a dial-up modem -- the Bell 103, which provided a 300 bps connection. (This model, incidentally, was for lease only.) In 1978, Hayes Microcomputer started selling a 300 bps modem, for $695. Today, a new home computer comes with at least a 33.6 Kbps modem, and more likely with a 56 Kbps modem. While this sounds like progress, it's actually rather pathetic.

For decades now, computer processing power and memory capacity have been improving according to Moore's law, which holds that memory capacity should double every 18 months, that is, improve 60 percent each year. If we had similar progress in data communications, 3 Mbps data connections would be normal for home users today.

What's the problem? It's just that until very recently, local access was a regulated monopoly. Each residence had, at most, one telephone service provider and one cable TV provider. But now change is in the air. Although political, regulatory, and legal issues still dominate over purely technical or monetary issues, competition is emerging. And politicians realize that many of their constituents take Internet connectivity very seriously.

Over the next three years, the battle for dominance in the Internet access market will be fought between the phone companies, cable companies, and entrepreneurs leveraging telephone or cable infrastructure. But, in the longer term, radio-based approaches will surpass them all.

INTERNET ACCESS: NEAR-TERM PROSPECTS
For most people today, the only affordable Internet access is a dial-up modem at a maximum of 56 Kbps per telephone line. The 56 Kbps limit is a result of the equipment at the telephone company's central office which is used to terminate the twisted-pair copper wire from each home. The copper wire itself can carry more bandwidth, but the existing digital telephone network restricts the achievable bandwidth to no more than 64 Kbps. The irony is that, except for the local copper loop, the telephone network is digital. But it is based on a 64 Kbps digital standard originally selected because it was the most efficient way to digitize sound from a 19th century microphone using 1960-era transistors.

The immediate contenders for better bandwidth to the home are digital subscriber line (DSL) equipment and cable modems.

DSL provides digital service at data rates from 200 Kbps to more than 8 Mbps over existing copper pairs that have been rerouted from traditional telephone central office equipment to special DSL electronics. Currently, all of the major telephone companies are experimenting with DSL, and US West is actually offering service in some areas.

Cable modems reuse cable TV wiring. This coaxial cable has much more potential capacity than the twisted pair cable used by telephone companies -- potentially many hundreds of Mbps. The initial cable modem deployments provide 30 Mbps of digital bandwidth shared with the usual TV channels and then parcel this bandwidth out to individual subscribers over connections supporting speeds from 1.5 to a few Mbps each. Cable modem service is deployed in several major U.S. cities today and is potentially available to 10 million homes.

Of course, change comes slowly to a regulated monopoly. Telephony companies currently charge $500 to over $1,000 per month for T1 service (at 1.5 Mbps). With DSL, a one-time purchase of $1,000 worth of equipment could support 1.5 Mbps or more over existing telephone lines that cost $10-30 per month. So while DSL could open up enormous new markets, it will cut into existing T1 revenues. Initial RBOC reaction has been highly retrograde.

Until recently, most telephone companies offered bare copper circuits for alarm systems and the like. Enterprising entrepreneurs were using these alarm circuits to experiment with DSL technology. But now, these offerings are vanishing. In 1997, US West (the most DSL-aware RBOC) filed with state regulators to discontinue Local Area Data circuits (bare copper circuits) in every one of the fourteen states where they operate.

The best hopes for DSL are competition from the cable companies and the Telecommunications Act of 1996. While the Telecommunications Act of 1996 is by no means the last word in deregulation, it at least defines a new class of regulated company, a CLEC (Competitive Local Exchange Carrier), and it requires that the ILECs (Incumbent Local Exchange Carriers, that is, the RBOCs) unbundle elements of their current infrastructure (like copper wire) and lease them to the CLECs at negotiated rates. This act has prompted some brave entrepreneurs (and their lawyers) to form CLECs so they can provide DSL service over ILEC copper wires. New companies like COVAD (in Santa Clara, Califor-nia) and Transwire (in New York City) are qualifying as regulated CLECs and are rolling out DSL service today. Early experience is mixed. There are few standards. Many existing copper pairs are too long or of too poor a quality to handle the highest DSL rates. But deployments are beginning.

Deployment of cable modem solutions is further along, but it also has its problems. The existing cable plant is primarily one-way only, feeding analog TV signals into subscriber homes. In order to make cable systems suitable for Internet access, cable companies have to upgrade feeder cables to the neighborhoods and replace existing couplers and amplifiers and other equipment with two-way equipment. Until three years ago, cable companies were so undercapitalized that there was little chance of that happening. Now, as a result of high-profile investments by Bill Gates and others, the market valuation of the cable companies has increased. Capital is available, and upgrades are well underway. Systems that are cable modem ready now pass nearly 10 million homes, and several hundred thousand cable modems are in service. So, at this point, cable modem deployment is way ahead of DSL.

Technically, performance should be equivalent. Cable modems multiplex multiple subscribers onto one high-bandwidth pipe. DSL systems multiplex many subscribers together at the central site. In either case, the quality of service you get will be a function of the company you deal with, not the technology.

In a very rough sense, DSL deployments require an investment of about $100,000 at the central site and then $1,000 per subscriber. Cable modem deployments may require a bit more at the central site, but only $500-700 per subscriber. These costs fall as volumes increase. And, as standards emerge, there is the potential for subscribers to buy their own modems (at ever declining prices). If you consider just the $100,000 cost (per central site) to get started, together with slightly more than 20,000 telephone central offices in the United States, you come up with $2 billion. That's a lot of money for a single high-tech start up, but it's not a lot of money spread out over many companies and many years. We could see high-speed Internet access potentially available to 50 percent of the U.S. population within 3-5 years.

DSL and cable modem technologies offer fairly dramatic improvements over the current state of Internet access, but there is another solution on the horizon with even greater potential.

INTERNET ACCESS: LONG-TERM PROSPECTS
Long term, the answer to Internet connectivity is radio. There's tons of potential in wireless access. As Paul Baren pointed out the Marconi Centennial (Bologna, Italy, June 1995), radio spectrum is not the scarce resource government that regulators would have you believe. The useable radio spectrum has been growing exponentially for 100 years, and this growth in useable spectrum is attributable to technological improvements, the results of efforts on the part of inventors and entrepreneurs. (The first was Marconi, who in 1895 showed how you could have more than one radio transmitter operating at a time by using resonant tuning.)

Today, there are billions of radio transmitters in the world when you count cordless phones, garage door openers, etc. Yet much of the radio spectrum is empty much of the time. By using low power, frequency agile, packet radios and intelligent receivers, we have the potential to get thousand- and million-fold improvements in utilization of existing radio spectrum. The issues are improved electronics (which are coming, in accordance with Moore's law) and government permission. Right now, there are companies matching available technology to each potential regulatory opening. All we really need are a few more regulatory openings.

The most recent opening is in local multipoint distribution services (LMDS) using a block of spectrum at 28 GHz that the FCC started auctioning in February. But the most technically interesting approach uses an unlicensed band (902-928 MHz) and small spread-spectrum radios to build a distributed, self-organizing, multihop wireless network.

At least two companies are entering this business, Rooftop Communications and Metricom, Inc. In the systems developed by these companies, each radio includes a router that runs protocols that route traffic hop-by-hop through as many radio sites as needed.

In the Rooftop Communications scheme, a group of radios share the same radio spectrum. Each radio broadcasts with only enough energy to reach a couple of nearby radios. As more radio/router combinations are added, the total system's capacity to pass messages expands. By reducing radio power so individual broadcasts only reach other radios in the immediate vicinity, it is possible to continue to in-fill additional radios reusing the same radio spectrum over and over again.

Assume each subscriber buys a radio/router and puts it on their roof. As more and more subscribers join, the web of connectivity in an area gets more dense and the traffic handling capacity goes up. Of course, one still has to pay an Internet Service Provider (ISP) who makes the connection for the radio network into the Internet backbone, but ISPs are by now a well-understood business.

Ten years from now, radio will be the method of choice for broadband connectivity. Fiber will come into a neighborhood or into an apartment building, connect to a radio base station, and everything else will be wireless. This neatly avoids the problem of dealing with the cable company or the RBOC. Subscribers just have to buy a radio/router and get immediately connected to their neighbors. They will have to pay an ISP for Internet access, but they will have a choice and not be limited by a monopolistic entity such as the RBOC or cable company that owns the cable running into their homes.

While radio technology is relatively expensive today, the price of electronics continually falls. And even with continuing government regulation of radio spectrum, the unlicensed bands provide an opening for wireless Internet access. But this is a ten-year prediction. For now, take whatever service passes your home, knowing that it can only get better in the future.

Brough Turner is senior vice president of technology at Natural MicroSystems, a leading provider of hardware and software technologies for developers of high-value telecommunications solutions. For more information, call Natural MicroSystems at 508-620-9300 or visit the company's Web site at www.nmss.com. E-mail to the author is also welcome.