Personal communication services show that it is possible to provide mobile access and have been popular among users. Now third-generation, or 3G, communications open the possibility of faster data rate mobile applications, but the spectrum needed for 3G overlaps some of the spectrum currently used for fixed wireless applications.
So it is time to take advantage of the lessons learned from existing commercial system deployments of cellular phone and existing mobile data access systems, such as Ricochet, cellular digital packet data (CDPD) and RAM Mobile Data. Taking advantage of innovative technology while concentrating on sound RF principles will provide 3G users with the best service combination of functionality and cost and will also encourage optimal use of a limited resource: spectrum at easy-to-use frequencies.
Fixed wireless will always be more spectrum-efficient than mobile access techniques because mobility requires an omnidirectional antenna (one that radiates in all directions). Mobility is possible only with relatively slow data speeds or with a very extensive infrastructure. It is technically and financially unfeasible to support high data rates in a mass-deployed mobile system. Of course, "high speed" and "mass deployed" are in the eye of the beholder and, more important, in this case may be traded off for each other.
To be sure, the exponential growth of the number of Internet users has been amazing; but even more astounding is the exponential growth of Internet traffic. This divergence between the number of users and the amount of traffic is the result of average users needing more bandwidth, that is, higher access speed. More bandwidth is required because more and better information, as well as more multimedia content, are available through the Internet. The better the Internet gets, the more of it everyone wants.
People will always want portability and mobility: the more convenience wireless engineers give them, the more they want. These two market realities are forcing the matter of providing high-speed access in mobile systems. Major telecommunications operators have spent billions and are preparing to spend much more to solve the last-mile conundrum. However, the simultaneous physical implementation of both high speed and mobility for a reasonable price is not so simple.
High data rates require broad bandwidths-simple enough. This was proved more than 50 years ago and is still true in the new millennium. The 3G mobile cellular phone systems offer Internet access with the promise of 2-Mbit/second data rates. Although these data rates are achieved quite handily in the wired world and routinely in the fixed wireless spectrum-for example, multichannel multipoint distribution service is 2.1 to 2.7 GHz and local multipoint distribution service, 24 to 40 GHz-they are a bit harder to achieve in the architecture of a mobile system. This is one reason that the 3G IMT-2000 proposal limits mobile users to 384 kbits/s.
System planners as well as politicians and FCC commissioners must be cognizant of the trade-offs involved in high-speed wireless systems. High data bandwidth comes either through unsophisticated modulation with wide spectrum allocations-which is very expensive-such as analog FM modulation, or through advanced, compressed modulation techniques such as 64 QAM (quadrature amplitude modulation) and its derivatives. More sophisticated modulations are used to reduce the required bandwidth for a given data rate. However, the more bits per hertz transmitted, the better the signal-to-noise ratio (SNR) required to allow accurate demodulation at acceptable bit error rates (BER).
At a given path loss, sufficient SNR is achieved either by narrowing the bandwidth, which limits the data bandwidth, or increasing the power of the transmitter. Of course, narrowing the spectrum bandwidth improves the noise power, but it also forces a higher-order modulation to transmit the same amount of data. Looked at in another direction, doubling the data rate may be achieved either by using a higher-order modulation with the same RF bandwidth-that is, going from quadrature phase shift keying to 16 QAM, which requires roughly 7 dB better SNR-or by doubling the RF spectrum bandwidth, which requires 3 dB more transmitter power.
The real-world realities mean that neither approach achieves the desired outcome of more data bandwidth at the same power level, since either the SNR must be increased to support the more complex modulation type or the received power must be higher to overcome the wider noise power.
Several solutions are available for high data rate systems. The transmitted power can be increased; the distance to the reception node can be shortened, resulting in an increased number of nodes, and/or directional antennas can be used so that the transmitted power is concentrated toward the intended receiver. All of those solutions will improve the signal level at the receiver. However, the first two result in increased background noise. On one hand, the users in adjacent cells also must increase their transmission power. On the other, they must, of course, be closer to adjacent basestations. Each of these approaches has a point of diminishing returns that limits net data throughput in an area because interference increases from nearby nodes. The transmitter power must be increased again to overcome this interference, or path length must be shortened even more to overcome the interference of other users in nearby cells.
That raises several questions. First, how do cell phones, with their omnidirectional antennas, operate with large cell sites? The data bandwidth is narrow, which means the SNR is improved by narrow spectrum bandwidths resulting in low-noise power bandwidths. The data rate for most cell phone systems is less than 20 kbits/s. Second, what about spread spectrum solutions? Deployment experience with code division multiple access has shown how difficult that is to do in reality. Spread spectrum is subject to the so-called waterfall effect when the traffic load approaches the upper limit. Once the number of users exceeds that limit, every user in a cell experiences quality problems. Next, what about existing mobile wireless data services? All of them have either very narrow data bandwidths or an extremely low user density. The data rates are advertised to be 70 kbits/s but experienced rates are typically less than half that.
Finally, how do fixed wireless systems operate with high data speeds? They overcome the path loss matter by using antenna gain to reduce wasted transmitted energy. The antennas provide directivity for the radiated energy concentrating it where it will enhance the SNR (and BER) rather than radiating it in all directions. Even a little antenna gain improves the situation considerably. An antenna with a 17-dBi gain at the subscriber site reduces transmitted power by a factor of 50 for the same SNR at the receive site. This means reducing the power from 1 W to 20 mW for a typical high data rate link. More important, the radiation is directed toward the intended receiver rather than omnidirectionally. The side lobes on a small directional antenna are typically more than 10 dB below the main beam, thereby reducing interference to other cell sites. Of course, mobility is nearly impossible with a directional antenna since the antenna must be aimed in the correct direction.
Another thing users want is an option for self-installation of the access module inside their homes. This means that the path loss calculations must take into consideration the attenuation that would occur through the building walls when wet from rain, or through a wall with aluminum-backed insulation or through a window screen. Those conditions make the argument for a directional antenna even stronger. Any transmitter would need to increase the output power enough to overcome the additional losses; therefore, the side lobes of a directional antenna would go up proportionally. However, they would still be lower, by the gain of the antenna, than if an omnidirectional antenna were being used.
The 2.1-GHz band is more vulnerable to foliage attenuation and losses through walls than are lower-frequency systems. Every experienced cell phone user has noticed how much better the lower frequencies penetrate a building-for example, broadcast radio, TV signals or even 870-MHz digital AMPS cell phones-vs. 1.8-GHz PCS phones. Generally, the lower the frequency the easier it is to receive a good signal or transmit from inside a building. Using a high-speed wideband access system at 2.1 GHz as proposed by 3G will be frustrating to users when inside a building and, once again, the distance to the access nodes will have to be reduced, resulting in more nodes and a high noise floor of interference.
Desired connectivity speeds will continue to increase, putting more demands on an access system. Any given connection speed will be less satisfying to users over time. It is easier to offer faster speeds to more users with directional antennas than with omnidirectional ones.
Everyone wants faster connectivity. Mobile access is an important business opportunity with many possible applications. However, the majority of legitimate mobile applications do not require high-speed access. For example, checking traffic, receiving financial updates or finding phone numbers, vendor locations, product price and availability and airline schedules do not require high-speed connections.
Going back to the principles outlined in the beginning of this article, the optimal solution is to use spectrum efficiently consistent with subscribers' requirements. One solution is to deploy high-speed, fixed wireless access using a large number of micro- and picocells. The microcells are wireless nodes that bridge a central community basestation and several neighborhoods' piconodes. For example, microcells might address distances of up to 4 km. A layer of picocell wireless bridges would support path lengths of 400 m. Those micro- and piconodes could be distributed based on three criteria: number of subscribers supported, maximum distance to the farthest subscriber and total traffic in the micro- and picocell.
In this scenario, most fixed or transportable subscribers would be able to find a piconode that has visibility from any room in their homes from which they wish to operate. The microcells could be co-located with existing cell phone hubs and the picocells could be roof-mounted nodes, each with a directional antenna toward the microcell and four sector antennas to cover the neighborhood's users.
This scenario allows very-high-frequency reuse through a large number of micro- and picocells that may be accessed via small directional antennas that are user installed. It takes advantage of the directional antennas to maintain low amounts of useless, randomly radiated energy.
Another major advantage of this multilayered concept is that truly mobile users may use omnidirectional antennas and work from picocells that are nearby. Of course, the system operator will want to limit the amount of traffic transmitted over the omni antennas to prevent the noise floor from rising. That could be done via differential pricing or restricting the upstream data rate when using an omni antenna.