What's Right With Telecom

This is part of IEEE Spectrum's special report: What's Wrong—What's Next: 2003 Technology Forecast & Review.

You know that quote about really advanced technology being indistinguishable from magic?" a friend recently wrote me. "I'm sitting here on my front porch, watching the rain, listening to Don Giovanni on my iPod, and sending this e-mail on my laptop, my cellphone by my side." It wasn't that long ago that this 57-year-old Philadelphia clinical psychologist was dialing a rotary telephone and playing vinyl records on a phonograph player.

Like my friend, we've all come a long way. Undoubtedly, any picture gallery of the telecommunications industry must display some disturbing images, including ones of the dot-com disaster, the collapse of demand for products and services, misguided regulations, accounting scandals, layoffs, and bankruptcies on a massive scale. But there's also the idyllic still-life-with-end-user painted above, featuring, as it does, what's still right in telecom.

Broadband Internet access continues to grow almost everywhere in the world, with especially rapid expansion in South and East Asia and Eastern Europe. Wireless local-area networks, most of them based on the IEEE's 802.11b standard, have popped up in transportation terminals, offices, coffee shops, and homes—including my friend's. New cellphone services such as text messaging and the sending of photos have been a big hit. Indeed, consumers have shown themselves willing to spend more than ever before on media and communications hardware and services. The challenge now is for companies to come up with offerings that will keep people digging into their pockets.

And they are. Higher-speed wireless services are in the offing: both faster 802.11 standards and the much-maligned third-generation (3G) cellular systems, which, despite uninspiring debuts in cellphone-friendly Japan, are poised for a much better reception in coming months there and in places like China and the United States. And coming not much later will be true broadband wireless networks—pure Internet-protocol (IP) voice-capable 2-Mb/s-and-higher data networks for which some analysts are even trotting out the term 4G. Connected to an Internet whose last-mile tentacles into the home are increasingly broadband—DSL, cable, and, eventually, fiber optics—the new wireless systems will give people new mobility and freedom not felt since the advent of the automobile in the early 20th century.

Going digital has already rewritten the business playbooks of telecommunications and the entertainment industry. Now, with combination cellphone/personal digital assistants (PDAs) and laptops and tablets equipped to commune with whichever connection is handiest, be it broadband data lines here, an 802.11 node there, a faster-than-narrowband 3G system everywhere, ubiquitous Internet-based communications networks will remap lives and lifestyles in the same way that freeways and autobahns did 50 years ago.

Broadband: alive and well, thank you

When added together, these elements of future networks will let us do remarkable things. But to read some recent accounts, you might think they've already failed. Take broadband: Bill Gates and countless other industry bigwigs have complained about slow adoption rates in the United States. Yet nearly 20 million U.S. households have signed up with broadband in the four years it has been widely available—a much faster rate of adoption than that of cellphones in the early 1990s (a common benchmark for rapid technology adoption). That's about 15 percent of all households. For comparison, only about 30 percent have one or more cellphones, despite that technology's 10-year head start. The research firm Technology Futures Inc. (Austin, Texas) says that broadband will hit that 30 percent mark in the United States sometime in 2004.

Even today, though smaller in number, broadband users account for more than half of all home-based minutes spent on-line. And outside the United States, the figures are even more favorable. In South Korea and Singapore, for example, more than half of all on-line households have a broadband connection.

Broadband providers are also working to improve the last-mile connections to the customer's premises. The trick is to avoid radical and expensive alterations to the central networks themselves. New technologies for squeezing more speed out of existing copper wire and coaxial cables are moving from testbed to roadbed.

For copper, two unheralded new specifications by the International Telecommunication Union (Geneva, Switzerland) would increase the speed and reach of asynchronous digital subscriber lines. (The specs, ITU G.992.3 and G.992.4, would provide other benefits as well, such as quicker start-up, and lower-power operation.) And for coaxial cable, a pair of improvements are in the works. One is a plan to use spectrum above 1 GHz, which broadband cable service currently does not, greatly increasing data rates. The other would boost capacity of cable systems by injecting ultrawideband signals into existing cable networks, again improving data rates.

Ultimately, it's all leading up to fiber-optic lines that run right into your house or small business. And it could be coming sooner than you think. The reason is Moore's Law, the doubling of computational power every 18 months, which may be ruinous to telecommunications business models [see "The End of the Middle"] but will be a boon to end users for whom broadband is still not fast enough or is just too expensive.

New developments in optical components promise to make it a lot less expensive to use fiber-optic cable to connect the Internet backbone to neighborhood-level network servers and routers [see "Cutting the Cost of Optical Components," and "Paving the Last Mile With Glass," IEEE Spectrum, December, [pp. 13-14]. Arguably, once these head-end stations are connected, it will be a relatively simple matter to extend the fiber out to individual homes, since telephone companies replace at least 3 percent of their last-mile copper every year, and many new housing units (1.5 million annually in the United States) could start with fiber.

Mobile, and loving it

Wireless digital voice and data services like text messaging are the other key market segments often left for dead by business commentators. In fact, for companies like Sonera, Vodaphone, and Sprint, they are the only bright spots in an otherwise downcast tech economy. Through much of Europe and Asia, 70 percent or more of all phone users have a cellphone. In data-besotted Japan, three-fourths of them use NTT DoCoMo's i-Mode or another wireless Web service.

Indeed, one of the main barriers to adoption of 3G cellular services has been the spectacular success of 2G data services, especially text messaging, which for Vodaphone accounted for almost 10 percent of all revenues last year and over 20 percent of all profits. News, weather, and sports scores, multiperson mobile games, music, e-mail, and even personalized ring tones are all big—and growing—revenue sources. Lately, as subscriber levels for plain old cellular phone service approach market saturation, these additional services are where the money is.

The reason is that these data services are usually "on the meter": phone companies charge customers per message, per megabyte, or per minute, so revenue goes up with usage. That's less and less the case for traditional local, regional, and even long-distance landline telephone service, where flat-rate fees for hundreds of minutes of talk time per month are common and even unlimited-usage plans are now being offered.

If each is good, both are better

If there's a killer app in telecom's future, it might be the marriage of broadband and cellular. Arguably their courtship began with Ricochet, a service that briefly disappeared when start-up Metricom Inc. went bankrupt in 2001, only to be resurrected by Denver-based Aerie Networks Inc. [See "What Went Wrong at Ricochet?," Spectrum, March 2002, pp. 60-61.] The network offers 128-171-kb/s wireless access, lower by far than broadband's 500-2000 kb/s, but triple the 40-70 kb/s that carriers like Sprint are (misleadingly) calling 3G. Ricochet is once again up and running in Denver and San Diego, and is expected in additional cities shortly.

Further in the future are true broadband cellular networks. A pair of companies are working on them, seeking to avoid what they see as the principal mistake of the 3G approach, namely, building a system in which data services are carried on top of what is essentially an architecture optimized to carry voice. Instead, working independently of each other and using different technologies, the two companies, Flarion Technologies Inc. (Bedminster, N.J.) and Soma Networks Inc. (San Francisco), have designed their equipment specifically to transport Internet data, which comes not in steady streams of fixed-length packets but in bursts of packets of widely varying size. Their networks work with IP.

Using IP from the ground up is important, according to Rajiv Laroia, founder and chief technology officer at Flarion, because in any well-designed wide-area wireless network, the most expensive part is the air interface, meaning the link that connects the end user to the wired part of the network by radio. It therefore makes sense to utilize that link as fully as possible—paradoxically, to make it the bottleneck in the network. Inevitably, therefore, frames (parts of packets) of data will be dropped in the air link. In traditional voice-oriented systems used for data, if heavy traffic requires 10 frames to be dropped, they probably come from 10 packets, and all 10 would have to be retransmitted.

In a well-designed IP-based wireless system, on the other hand, an IP router in the base station closest to the user would know which frames are in which packets, and would choose to drop the 10 frames from a single packet; only one packet would have to be retransmitted. (In fact, it would drop the entire packet and maybe make it unnecessary to drop any more frames.) The result: higher overall throughput, which is what broadband is all about. And by using a variety of so-called quality-of-service schemes, which assign priorities to packets depending on how much delay they can tolerate, IP networks can be much more effective in carrying voice traffic than voice-centric networks are at carrying data.

Flarion has a test cellular network up and running at its headquarters using a proprietary variation of orthogonal frequency-division multiplexing, which the company calls flash-OFDM. With OFDM, the spectrum assigned to a session is divided into a series of tones, and the transmission hops among the tones, making 10 000 hops per second. Interference with other users, both within a given cell and among cells, is thereby reduced, and the number of users for a given piece of spectrum is increased.

The same peer-to-peer architecture that made music sharing such a success can work for data transmission

Like Flarion, Soma uses a variety of techniques, such as adaptive modulation and adaptive error correction, to improve cell coverage and increase data rates. But instead of OFDM, it uses code-division multiple access (CDMA) as the air interface to cellular networks. With CDMA, each transmitted data stream is combined with a pseudorandom code that makes it appear like noise to other sessions in the same cell that are using different codes for their communication.

The scheme works much as at a cocktail party where every conversation takes place in a different language (code). To people who speak only, say, English, a nearby conversation in French simply adds to the background noise. Provided that the noise doesn't get too loud (and CDMA systems take great care to control power to prevent just that), the technique allows multiple sessions to occupy the same spectrum in the same space at the same time.

"Whether you start with OFDM or CDMA is less important than these details of implementation," says Greg Caltabiano, a senior vice president at Soma. Still, the company thinks there's an advantage to staying within the CDMA ecosystem, as Caltabiano calls it. "You can use a lot of the same test equipment, chip sets, algorithms, and deployment modeling software developed for conventional voice-based CDMA systems," he says.

There is also a non-3G-but-not-exactly-cellular, pure-IP data network architecture. It's a so-called mesh network, in which each user's device (laptop, PDA, or cellphone) is also a transponder, receiving and then retransmitting the packets of other users. In other words, the same fundamental peer-to-peer architecture that made music sharing such a success in the past few years can work for data transmission itself. Remarkably, such networks can keep adding new users long after a traditional cellular system would have to stop, because each new user increases the network's capabilities.

In some such networks, the nodes are fixed locations like homes and offices [see "Wireless Broadband in a Box," Spectrum, June 2002, pp. 38-43]. But one, that of MeshNetworks Inc. (Maitland, Fla.), is unique, in that its nodes are laptops, PDAs, and even phones that can be traveling at highway speeds.

MeshNetworks' products are dual use: they work in their own proprietary network and are also compatible with IEEE 802.11. Indeed, it would be foolish to not capitalize on that standard's phenomenal popularity, another of the industry's unalloyed silver linings.

Analysts at Frost & Sullivan (New York City) say that in 2002, companies and home users bought an astonishing 14.4 million wireless local-area-network (WLAN) routers, most of them for the 802.11b standard. These are the local-access points in cafes, businesses, and airports that serve as the nodes to connect appropriately equipped mobile computers with wired broadband networks. Just last month, Intel, AT&T, and IBM announced the formation of a start-up company, Cometa Networks (New York City), to place thousands of new access points in just such locations throughout the 50 largest U.S. metropolitan areas.

802.11's potential is in fact great enough that some observers fear it will affect the ability of cellular providers to raise capital, needed for upcoming 3G auctions. Frost & Sullivan predicts that annual sales of WLAN products will rise to over 36 million by the end of the decade.

By then, we'll also have much faster flavors of 802.11. The b version of 802.11 will soon give way to the a type. The a version uses unlicensed spectrum in the neighborhood of 5 GHz and tops out at 54 Mb/s, almost five times as fast as 802.11b. Motorola's Canopy access point device is one such product.

The race for ever-higher data rates recently drew in Lucent spinoff Agere Systems Inc. (Allentown, Pa.) and the Eindhoven University of Technology in the Netherlands. They demonstrated a chip set that joins three 802.11a antennas to create a single network connection of 162 Mb/s across a distance of about 10 meters. When this is combined with improvements in digital encoding, you could download an entire DVD-quality television show in the same time it takes to grab an MP3 song today—and do it while waiting in line for a double mocha cappuccino at your local coffee shop.

All that speed would be great, but what would be really useful today is a unit that acts as a cordless phone when you're close to your office, switches to a WLAN elsewhere on the corporate campus, and passes seamlessly to a broadband cellular network when you move outdoors. Another Lucent spinoff, Avaya Inc. (Basking Ridge, N.J.), is working on just that. Indeed, the shelves in research larders at companies like Avaya, Agere, Motorola, Flarion, Soma, and MeshNetworks are already stocked with wonderful new technologies. The question is whether these firms can hang on until a global economic recovery opens up corporate checkbooks.

If you build it, killer apps will come

Reliable high-speed mobile Internet access will be a foundation for products and services we can barely imagine. Only when the networks are built will the lightning bolt of invention strike clever engineers and entrepreneurs. But we already know that our increasingly mobile lifestyles will make cellphones and PDAs even more important than they are today.

One thing that will make cellphones truly indispensable at last is an application that turns a handset into a kind of wallet. At least that's the vision of Derick Wilson, chief executive officer of Optical Antenna Solutions (Nottingham, UK). His company has a patent on a light-concentrating lens originally developed at the University of Warwick. The lens allows the infrared beam found in PDAs and some cellphones to transmit across longer distances, even with not very accurate aiming.

According to Wilson, this is the missing piece of the puzzle for systems that let you point your cellphone and pay—at fast-food drive-through windows, supermarkets and department stores, vending machines, and even for that cappuccino at the coffee shop. Wilson is already talking with credit card companies like Visa. By storing credit card information on the phone itself (suitably encrypted, of course) and using the same sort of personal identification numbers used at automated teller machines, card companies hope to put a big dent in an annual credit-card fraud tab of $17 billion.

Another e-mail from my friend, sent from her porch, says it all: she's getting a Sony Ericsson T68i phone/personal digital assistant. "Finally, I can synchronize the data on my desktop, my laptop, my iPod, my cellphone, and my PDA." Magic.