June 2000

BY FRANK KELLY

The battle for consumers' hearts and minds is being fought in the telecom sector over the last-mile connection -- the so-called local loop. While long-distance has been parceled out by market forces among a few large and many small competitors, the copper wire connection between your home and the telephone company's central office today is firmly monopolized by regional and local telephone companies.

AT&T has responded to this challenge by spending upwards of $100 billion to acquire cable TV companies, the only other option that already offers the complete bundle of services that will grow their volume and profit. AT&T, MCI WorldCom, and Sprint see clearly that without owning this crucial link in the telecom chain they cannot provide a "wire" into the home. More or less as a response to AT&T, MCI and Sprint in turn invested in an almost forgotten technology with equal potential and a few advantages. The Sprint and MCI bets are on Multichannel Multipoint Distribution Service (MMDS). MMDS was instituted in the early 80s, when it became clear that TV viewers wanted more and better video, and that stringing coaxial cable to every home would be increasingly costly as new areas to be served were less and less dense.

Also called wireless cable, MMDS video is a simple technology, a variant on the original community television concept. An MMDS headend translates local and distant television channels up in frequency to a portion of the microwave spectrum at 2.5�2.7 GHz. These signals are typically radiated omnidirectionally at modest power from a tall tower more or less in the center of town, and are received with an inexpensive 18" parabolic dish and block downconverter. The subscriber's own TV set performs tuning and demodulation. In many areas, this relatively low-tech solution can serve a 30-mile radius, and provide improved television service to hundreds or thousands of subscribers who couldn't reasonably expect cable connections any time soon.

The FCC licenses just one MMDS provider in any given area. Each over-the-air TV channel occupies six MHz within the MMDS band, and so the service was originally limited to providing no more than 31 channels* of television. MMDS flourished through the mid-nineties, but as cable extended its reach and offered many more channels, MMDS operators found it difficult to compete. The addition of digital video compression allows an MMDS system to transmit up to 150 channels, but this requires a significant investment at the headend (on the order of $20,000�$40,000 per channel), and a digital set-top box in each subscriber's home. Although there are a number of digital MMDS options in operation today in the U.S., digital MMDS came too late for most MMDS licenses that were facing bankruptcy at the end of the 90s.

THE VOICE/DATA CONNECTION
The Sprint and MCI entrance into this picture came with startling speed. In April of 1999, these two giants, acting independently, spent a total of nearly $1.9 billion dollars to acquire the companies that held close to 50 percent of US MMDS licenses. The catalyst for the buying spree was a change of the rules of the game by the FCC.

Starting in 1996, an industry group dubbed the Ticket Two Task Force lobbied the FCC to allow two-way use of MMDS frequencies. In September of 1998 the FCC issued a report and order permitting two-way MMDS. Thus it became possible to use MMDS to provide a wide-bandwidth, two-way link between individual subscribers and the MMDS headend. The formerly monopolized local loop could be bypassed without trenching down every street and road to lay new wires or cable. At the same time, the explosive growth of the Internet was fueling demand for a wide bandwidth local loop, as more and more consumers searched for a way to speed their Web experience.

In the pursuit of lower CPE costs, MMDS has benefited from technology and manufacturing capacity originally put in place for cable video services. By using a well-documented standard called DOCSIS** developed for cable modems, two-way MMDS benefits from the economies of scale achieved in producing cable modem chip sets. It was originally assumed that existing design cable modems could be plugged into a wireless cable system to serve this new class of users. This turned out not to be true. The wireless environment has noise, fading, and dynamic range issues not encountered at the end of a coaxial cable. It was determined that the customer's modem should be designed to deal with wider frequency variations to allow drift and temperature variation specs to be relaxed in the design of CPE (customer premises equipment) for lower cost. Also, incorporating a selection of modulation methods (QPSK, 16QAM, 64QAM) would allow extensions to the range of a wireless data system. By enhancing the DOCSIS standard with wider frequency acquisition and dynamic range limits, and the additional modulation densities, so called DOCSIS+ modems that still use the low-cost DOCSIS chip sets have been designed for wireless.

The technology of broadband wireless systems has rapidly become quite sophisticated. Latest generation transmitter equipment offers system designers low phase noise transmitters (for lowest bit error rate), a selection of modulation formats, and frequency-reuse cellularization schemes to serve large populations of users within the 200 MHz MMDS band.

Today, typically, a two-way MMDS system will transmit from the headend up to 27 Mbps in a single six MHz-wide MMDS channel (other channels remain available to serve video customers). Because individual users share this spectrum in bursts, it is possible for each of hundreds of subscribers to see an effective data rate of approximately one Mbps using that single channel. Because Internet use is largely asymmetrical, with much more data flowing "downstream" (from the headend to the user) than in the reverse direction, a much smaller bandwidth can be used in the upstream (back to the MMDS headend). Therefore, a very low power, relatively narrowband transmitter is used at the customer's premises. For the upstream link, typically an 800 KHz-wide channel is modulated with QPSK data. Shared among many users, this provides effective upstream data rates of at least 64 Kbps to each subscriber.

By carefully optimizing the upstream and downstream link budgets, a system designer can configure a system to serve 1,000 to 2,000 initial subscribers within a 20�30 mile radius of a tower using this single six MHz channel. This startup scenario is called a supercell. It initiates high-speed Internet service to business and residential subscribers of an entire city and its suburbs at modest cost. As the subscriber list grows, more MMDS channels can be used, and sectorized antennas at the hub allow for frequency reuse. As a further buildout strategy, "minicells" are strategically placed to serve pockets of higher subscriber density. As minicells are usually within the supercell coverage, they need to use different frequencies than those of the supercell to avoid interference, and so that sharing the bandwidth does not restrict individual subscribers' data rates. Thus the 200 MHz bandwidth of the MMDS band does not become an issue. It is possible to show that an MMDS system using a simple cellularization plan can serve a million or more subscribers in a densely populated metropolitan area.

A GROWTH OPPORTUNITY
The technology is versatile. Given a high data rate digital connection, carriers are not limiting themselves to Internet access. Voice-over-IP additions to the transmission standards and to equipment designs will soon add four or more POTS telephone circuits to the menu of offerings of MMDS. What then are the strengths and limitation of MMDS as a broadband alternative to the local loop? The major strengths of MMDS derive from its wireless technology.

The low microwave MMDS band is unaffected by rain and other precipitation so MMDS casts a wide net. Service can be built in a city very quickly by installing a single supercell. This gives MMDS a time-to-market advantage over other broadband services. As recent history has shown, upgrading a city's cable plant or its telephone infrastructure for DSL can take many months, even years. There may also be significant infrastructure cost advantages in particular markets, depending on the age and quality of the existing wire/cable plant. Beyond the initial cost, MMDS' ability to first serve a wide area of sparsely placed initial users, then grow with the subscriber base in a non-obsoleting manner, can be an important economic and marketing plus. Perhaps most important to some carriers is a political benefit. MMDS has the ability to bridge the "digital divide," to "serve the underserved"-- those rural and outer suburban users of the Internet who have no hope in the short or even medium term of getting broadband access.

At the same time, MMDS' limitations are real. The MMDS spectrum is only 200 MHz, and therefore it is not expected to provide the huge bandwidth connections -- the "fat pipes" --required by very large businesses. However, the frequency reuse we have discussed allows this modest slice of spectrum to serve great numbers of small to medium-sized businesses, and residential customers. There are line-of-sight issues to be resolved as well, but cellularization and some special purpose modulation techniques can be deployed in shadowed areas and overcome this problem.

Given these factors, where is wireless broadband access today? Sprint, MCI WorldCom, and BellSouth, another large holder of licenses, have announced either large-scale trials or phased deployments of the technology before the end of 2000. Equipment suppliers already offer carrier-class headend equipment, and costs of the customer premises equipment are being driven down by projected large production quantities. The sub-$500 set of modem and antenna-transverter is said to be well within reach. Although MMDS broadband wireless has only been deployed in a few small cities as this article is written, Gartner Consulting estimates that at least 20 percent of the 220 million US households will still not be able to get wireline broadband access four years from now. This represents a huge group of bandwidth-hungry consumers that MMDS-based broadband access can serve well.

*The ITFS channels were originally reserved for educational use in the U.S. These channels are often used commercially in other countries, and have been leased for commercial use in many localities in the U.S. as well.

**"Data Over Cable Service Interface Standard" issued by the Multimedia Cable Network System consortium.

Frank Kelly is director of marketing operations for ADC Telecommunication's Broadband Wireless Group. ADC is a global supplier of transmission and networking systems for telecommunications, cable television, broadcast, wireless, and enterprise networks. For more information, visit ADC's Web site at www.adc.com.