New technologies become encrusted with conventional wisdom surprisingly quickly, and broadband fixed wireless, sometimes known as broadband wireless access (BWA), is no exception. Here then is a list of "true facts" about broadband wireless that turn out to be not exactly correct on closer examination:
DSL and cable modems are the only two broadband choices. Cable and digital subscriber lines are providing millions of people with fast Internet connections and "always-on" service. But DSL has a limitation of wireline distance from the telephone central office, and data over cable requires the cable company to spend substantial sums to upgrade its outside plant. Businesses often find that cable does not pass the building, especially in industrial parks in the suburbs. This leaves a large fraction of the population (some say it won't be lower than 20 percent anytime soon) without broadband service.
Broadband fixed wireless is a third alternative. It uses microwave transmissions to deliver a broadband connection to a customer's home via a small roof-mounted antenna. A central hub serves many subscribers on a point-to-multipoint basis.
BWA is relatively new. In September 1998 the Federal Communications Commission first allowed the multichannel, multipoint distribution service band to be used for two-way data delivery. Commercial rollout of this service only began in the second half of last year. Sprint and WorldCom spent close to $2 billion collecting MMDS city licenses after the FCC's decision. They have announced plans to provide service in scores of U.S. cities by the end of 2001.
Thus there is hope for those who can't now, and may never, get a cable or DSL connection. By the end of 2001, about a quarter of the U.S. population will be covered by the BWA alternative.
All wireless Internet is the same. Recently there has been heavy press coverage of the possibility of wireless Internet via mobile phones. Shorthand for this technology is 3G (third generation), or 2.5 G. This is often said to be broadband, since data rates of 384 kbits/second or more are projected. However, the demand for spectrum and the signal-to-noise ratio margin required for user mobility still limit this technology to relatively low data rates.
Broadband fixed wireless is an entirely different technology. Because the subscriber is in a fixed location with respect to the hub, the link can be optimized and much greater bandwidth and data rates can be offered.
There is a good reason why fixed wireless will always be able to provide faster connections than mobile wireless. Shannon's Law says that the maximum theoretical data rate in a channel is proportional to S/N. A link in which the two ends are fixed will always have an S/N advantage and, therefore, a data rate advantage. Virtually all the conceivable optimizations that can be applied to mobile links can also be applied to fixed links, which will allow fixed wireless to maintain its advantage.
All broadband fixed wireless is LMDS, at millimeter-wave frequencies. Local multipoint distribution service is a millimeter-wave service with huge bandwidth-more than 1 GHz-but it suffers from rain attenuation (fade) and high radio costs. Recently two large LMDS service providers entered bankruptcy proceedings, and some of the press asked whether all of broadband fixed wireless was in trouble. But in addition to LMDS, BWA can use the MMDS band, as well as bands at 2.4 GHz and 5.8 GHz that do not require licenses. These three kinds of BWA--LMDS, MMDS and unlicensed band-have greatly different capabilities, costs and business models.
MMDS-band service includes operations in the 2.1-GHz MDS band, the 2.3-GHz wireless-communications service band and the 2.5- to 2.7-GHz MMDS/instructional TV fixed-service band. Those are licensed frequencies; for any frequency channel there is one and only one service provider that has exclusive use of that channel. For that reason WorldCom and Sprint bought MMDS licenses rather than set up shop in the adjacent unlicensed band.
The low microwave frequency offers very favorable propagation; there is no significant effect from rain or any other weather condition. License protection from interference and favorable propagation allow MMDS BWA systems to use supercells of 20 to 30 miles in radius. This is a strong advantage, because a whole metropolitan area can generally be covered by one supercell, leading to a virtually instant rollout of service everywhere in that market.
The third kind of BWA service uses unlicensed frequencies, avoiding the cost and delay of license procurement. This leads to the next misconception:
Unlicensed-band frequencies are free, with no penalty. Unlicensed-band BWA has grown rapidly because there is no license to obtain and because the lower frequencies have more favorable propagation than LMDS. However, the operator shares the band with all comers; there is no area in which it has exclusive use of the channel. Thus, unlicensed-band operation faces significant interference issues. Moreover, a competitive system may enter the area next month or next year and force today's operator to share the spectrum.
Sophisticated strategies for dealing with interference are provided by suppliers of unlicensed-band systems, but these tend to limit throughput, reduce cell size or both. For that reason, unlicensed systems have been more often deployed to serve a campus rather than a whole city.
Fixed-wireless signals can punch though walls with no penalty. This fallacy comes from the blurring of the distinction between fixed and mobile wireless. Certainly, mobile wireless telephony works inside buildings-why should fixed wireless require outdoor antennas? The answer lies in the differences between the two services: Wireless telephony deals with the path loss inherent in indoor operation by building in a large margin of signal-to-noise ratio, providing a very small bandwidth within a fairly small cell, of which there are hundreds in many large cities.
An "indoor" fixed-wireless system would have to make similar compromises: try to push up the data rate to something competitive with DSL, keep the cell size quite small and use peak power transmissions from the customer premise equipment (CPE) that may bump up against FCC RF-intensity safety limits. Within this set of trade-offs, it is possible to architect a system. Whether such a system's cost and performance are economically viable for large-scale operation remains to be proven.
Every subscriber needs nonline-of-sight (NLOS) operation.
Reading recent articles and press releases on broadband fixed wireless, you will soon gain the impression that nonline-of-sight operation is absolutely necessary, indeed the holy grail of this technology. True, serving 100 percent of potential subscribers in an area will require technology that does not depend on line of sight to the hub. What is overlooked is that techniques to overcome NLOS problems are far from cost-free. Not only do such technologies become more complex to minimize multipath, but also they need an increased signal level to overcome path loss, which is inevitable in NLOS situations. Technologies that minimize multipath and provide a large margin of signal level to overcome path loss will inevitably cost more, especially in the price of the customer premises equipment. (Low CPE price is the real holy grail of BWA.)
Also overlooked is that a large fraction of subscribers will have line of sight to a hub station. This is true when the hub is a supercell, with a tall tower in the middle of the service area, and more true if a service area is covered by a number of smaller cells. A subscriber who can't see hub A may have clear line of sight to hub B or C.
TDD is more spectrally efficient than FDD. On the face of it, this seems self-evident. Time-domain duplexing, in which the outbound and return data time share the same slice of spectrum, seems to be inherently more spectrum-economical than frequency-domain duplexing, where different RF channels are used for the BWA downstream and upstream signals.
In FDD you have to provide a guardband between transmit and receive, thus giving up spectral efficiency, while in TDD you must allocate a guard time between transmit and receive and thereby lose spectral efficiency. TDD is more spectrally efficient in the special case where only a single narrow channel is allocated, and duplex filters providing spectral efficiency are difficult to make. For MMDS and LMDS, where the allocated spectrum is larger than a single user demands, FDD is more efficient because a spectrum pair can be allocated with sufficient guardband and the guardband can be used in another cell.
A "truck roll" is to be avoided at all costs. There's a grain of truth here. In cable TV service, the fewer times that a technician must go to a subscriber's home (a truck roll), the better. Both the technician and the truck are large investments, so truck rolls can be costly, eating up system profits at a furious rate if there are too many of them.
However, in a newer technology like broadband wireless, the initial truck roll to install the antenna and modem should be seen as an opportunity. A properly trained technician will not only ensure that the customer starts with optimum performance from the system, but will also assist the customer (within reason) to configure his or her computer or home network. This is a not-to-be-missed chance to build customer satisfaction, which is absolutely crucial to avoiding that other dreaded demon of these services: customer churn.
Moore's Law will solve all cost issues. Moore's Law is the idea that the number of transistors in the latest microprocessor can be expected to double every 18 months. The corollary is that whatever computing power is required to solve a problem will be available next year and at reasonable cost. But all issues cannot be resolved by the addition of more computational power. The cost of digging up streets to install fiber or better cables isn't helped by cheaper computing power. Similarly, if RF power or regulatory problems arise, Moore's Law won't be the solution.
Fixed wireless is a short-term solution. Portability and vehicular mobility are key advantages of a wireless transmission compared with twisted pair, coaxial cable and optical fiber. Mobility is the one attribute that ensures radio waves will remain a viable medium indefinitely. Fixed wireless, on the other hand, could be obsolete when every home, business and school in the world has a high-speed wireline connection provided by every carrier wishing to compete. This is unlikely to ever happen, however-almost certainly not in the next 50 years.
Fixed wireless will have a role as long as it's costly to install and maintain wired connections to every potential customer. Ubiquitous wired connections have only been provided by carriers graced with monopoly status, which is now for practical purposes a thing of the past. In any case, monopolies tend to be slow to implement new technology, and they charge artificially high prices. In the case of broadband access, even monopoly status has not motivated carriers to provide ubiquitous service. This makes fixed wireless, with its fast buildout to cover an entire market, a viable medium for providing competition in the delivery of broadband access.