DATA OVER WIRELESS A PRIMER

By Matthew Oristano

Chairman and CEO, SpeedChoice

 

INTRODUCTION

 Up until as recently as 1996, the only economical way for a business to receive and transmit high-speed data (defined for our purposes as data received at speeds of one megabit per second and higher) was through the wired infrastructure built by the regional Bell operating companies (RBOC's). Aside from some small portions of very dense major metro areas that could support wired competition, there were no other options. And outside of businesses, of course, there were no options, period. Home data users had to content themselves with modems running at 28.8 Kbps (kilobits per second) over telco twisted pairs.

In the space of only three years, spurred in part by the growth of the Internet and in part by the opportunity of competing with an overpriced monopoly, several new wireless infrastructures for high-speed data have been proposed and are being built. This has the dual benefit of reducing costs for those businesses that bought high-speed data lines from RBOC's, and vastly extending the reach and coverage of data accessibility, to the point where for the first time, small businesses and residences can economically participate.

This explosion of High-Speed Access (HSA) will be the infrastructure supporting the second generation Internet, with higher levels of interactivity, media content, and computing power, than the Internet (first generation) now in use. The participants in this wireless explosion fall into certain broad categories that we will examine below. As you will see, the functionality and business plans are not the same. MMDS and LMDS, for instance differ greatly in cost and ultimate target markets. But in the end, an exploding HSA pie will ensure that each wireless data entrant, if well run, can have a profitable piece.
 

FIXED VS. MOBILE

 Up until recently, the mention of "the wireless communications industry" meant one thing; cellular telephone. That does not hold true anymore. Therefore, it is necessary to distinguish the sort of wireless data options we are talking about. The nation's cellular and PCS networks utilize highly cellularized architectures with low-density modulation schemes optimized for mobility . This means a great deal of system capacity is dedicated to ensuring that a moving vehicle can remain in contact. No matter how much ingenuity is applied to other uses of the network, the fact is that the original and highest priority use of it is to provide digitized mobile voice connections at only about a 10 Kbps equivalent data rate.

While there are initiatives underway by such companies as Microsoft, Qualcomm, and Motorola to overlay certain data capacities on such networks, they will still be configured for mobile or semi-mobile access using base stations with low cost omni-directional antennae. Cost-per-bit-per-second will likely be at a premium, and reflect the mobility aspect. The new wireless options are called "fixed wireless", as opposed to "mobile." Fixed here refers to a fixed location. In this case, the antenna is highly directional, high gain, and mounted to a building. Since buildings don't move, system capacity that would otherwise have been used to ensure uniform coverage for mobility can now be redirected to providing higher throughput.

Denser modulation schemes requiring higher signal-to-noise ratios can be used. And data rates can easily reach 10 megabits per second and higher, a 1,000-fold increase from current mobile capacity. As a result, cost-per-bit-per-second is orders of magnitude lower than in a mobile system. While mobile data access systems are being designed and will be built, our assumption here is that they will attack a fundamentally different market opportunity, and be some years behind their fixed counterparts with regard to wide-scale deployment. We will confine our attention to fixed wireless systems from here on.
 

WIRELESS FREQUENCY BANDS

 During the latter half of the 1990's, the FCC was under a congressional mandate to raise revenues through the auctioning of spectrum. In the process, it made several new bands of wireless spectrum available. At the same time, due to the Telecommunications Act of 1996, it was under a mandate to create viable competitive opportunities for wireless competition to RBOCs. This led it to enhance the capacity of certain preexisting spectrum licenses. The result is a host of new omni directional wireless HSA networks under construction. While there are other frequency bands (4.6 GHz, 12 GHz, etc.) where private users can provision individually licensed point-to-point data links at high cost, these will not be the source of the HSA explosion. Rather, the new allocations promote omni-directional transmission, with no receive site licensing required. We will focus on these new omni-directional transmission bands.

The MMDS Band (2.5 GHz)

During the 60's, 70's, and 80's, the FCC allocated approximately 200 megahertz of spectrum at 2.1 and 2.5-2.7 GHz frequency for television transmission. With the new digital technology and the new competitive mandate, the FCC greatly increased the flexibility of this band beyond simple video to full two way digital communications, excluding only mobility. In two separate rulemakings in 1995 and 1998, the FCC allowed for digital transmission utilizing CDMA, QPSK, VSB, and QAM modulation schemes yielding up to five bits-per-hertz (one gigabit-per-second total raw capacity for the band), and return transmission from multiple sites within a 35 mile radius protected service area. The spectrum is utilized in omni-directional fashion from a central antenna, and may be cellularized as necessary. As we will see below, a major advantage in using this band is that the physical need to cellularize is much less than in other major bands. With appropriate terrain characteristics, a single MMDS cell can cover a 35 mile radius, or 3,850 square miles. Current HSA systems using this band utilize cable modem technology, delivering 10 to 30 megabits-per-second (Mbps) downstream, and 32 Kbps to 10 Mbps upstream. CPE costs and coverage areas make this band appropriate for full coverage of large land areas, and provision of service to small businesses and homes not in dense clusters, as well as higher clustered businesses. Major deployments in this band have been begun by SpeedChoice in Phoenix and Detroit, and by Wavepath (Videotron) in San Francisco and Silicon Valley.
 

The DEMS Band (24 GHz)

This band was originally allocated at 18 GHz, with 100 MHz bandwidth. Teligent Corporation consolidated ownership of the band, and convinced the FCC to relocated it to 24 GHz with a 400 MHz allocation. Teligent is the only operator at this band. The frequency requires a high level of cellularization to cover a large area, and is particularly suitable for dense urban core markets. Cell size is about 2 miles in radius, or 12.6 square miles coverage area per cell. Teligent is deploying a wireless ATM backbone solution which is primarily geared to provide standard telephone service (POTS) at a 30% discount to RBOC prices. Along with this telephone service, Teligent can provision T-1 speed (1.544 Mbps) data links, but these links are not their primary business, and are bundled exclusively with POTS.
 

The LMDS Band (28 GHz)

This band was auctioned in 1998, with only a few major players participating. It consists of an "A" block with 1150 MHz bandwidth, and "B" block with 150 MHz bandwidth. With the purchase of WNP Communications by NextLink, the two major winners consolidated in 1999, and NextLink now owns 95% of the LMDS spectrum in the top 30 markets. Vendors have not finalized equipment specification or pricing, but early indications show a preference for ATM-based solutions similar to those employed by Teligent and Winstar. As with all solutions in this frequency range, a high degree of cellularization is required with this band. Cell size is about 2 miles in radius, or 12.6 square miles.
 

The 38 GHz Band

This band is primarily licensed to Winstar and Advanced Radio Telecommunications (ARTT). Winstar utilizes ATM-based equipment from Lucent, and provides POTS and high-speed data. The extremely high frequency used with this band requires intense cellularization, with cells of 1 mile radius, or 3.14 square miles. It should be noted that with all of the high-frequency bands, initial deployment in a city may not be omni-directional. It is more economical early on to deploy point-to-point links as customers are sold, creating a spider web-like network of buildings. Later, an omni-directional cell site can be overlaid if economies warrant. However, this will impose certain retrofit costs on the infrastructure.
 

Other bands

The FCC has licensed a variety of other bands for omni-directional transmission, but none with the bandwidth and exclusive licensing structures of those above. For instance, the WCS band was auctioned off in 1996. While this band at 2.3 GHz frequency would ordinarily have good operational prospects, its narrow allocation (20 MHz) and inopportune channel configuration make its use primarily attractive in conjunction with other bands. Also, there is public spectrum at 2.4 GHz called the ISM, or Instrument Scientific Medical band, which some ISP's have begun dabbling in for wireless links. Since this is public spectrum, there is no exclusive license, and so long as equipment meets FCC specs, anyone can operate in the band. This raises the specter of overcrowding, interference, and breaches of security. It is unlikely that a business would want to use this band for mission critical applications, and it is also unlikely that this band would scale appropriately for mass residential use.
 

SIGNAL PROPAGATION

 The single most important factor in the structure and economics of the infrastructure required for a given wireless frequency band is the band's propagation characteristics. That is, how far does the signal reach at a given power level under given terrain, foliage, and weather conditions. First, one variable is constant across all the spectrum bands discussed above: They require direct line of sight. This means the path between the transmitter and receive site must be substantially unobstructed. If terrain is hilly, or if foliage is dense, then line of sight opportunities must be multiplied through cellularization. This is true of all of these bands. If it is less true of cellular telephone and PCS, this is not because the propagation characteristics of these bands are different, but because they utilize a high degree of cellularization combined with extremely low density modulation schemes which do not require as robust a link, and therefore yield low data rates.

In dry weather, the primary variable determining signal propagation distance is the frequency band. Certain low frequency bands can usefully propagate for thousands of miles, and certain high frequency bands (optical bands, for instance) for only some thousands of feet. Beyond a certain distance, a receive antenna of given size and gain is unable to receive the signal. One could always put larger and larger receive antennae in place, but this obviously would severely limit the willingness of homeowners and businesses to use the service. Generally speaking, an antenna size of 18 inches or less is essential to a service for wide acceptability. Looking at the relative propagation characteristics of the bands listed above then, we see an ever-shorter transmission distance.

Thus, a signal at 38 GHz frequency will lose 230 times more signal strength over a one mile transmission path than an MMDS signal. A 28 GHz signal will lose 130 times more, and a 24 GHz signal 99 times more than an MMDS signal over a one mile path. These calculations are for dry air only, and the comparisons are further weighted in favor of MMDS when rain fade is accounted for. (See below.)At certain frequency bands, the problem of attenuation is greatly exacerbated by effects of weather. At high frequency bands such as 24, 28, and 38 GHz, wavelengths are short enough that raindrops can actually present line-of-sight obstacles, and greatly attenuate the signal. This then requires a further shortening of signal radius over what would otherwise be available in dry air. In order to engineer systems with five nines reliability or greater, transmission radii must be kept to a minimum at these bands. Even so, certain links may go down in real but statistically unlikely torrential downpours.

With regard to the MMDS band, due to its longer wavelength, rain represents no obstacle to a properly engineered system. Due to its prior incarnation transmitting TV signals, there are MMDS systems with years of continuous transmission history, and no effects whatsoever from rain. While there are engineering factors that can be applied to the higher bands to take account of rain, they have not been tested by real world commercial operation for years at a time. The design choices at the higher bands are essentially statistical "?bets' that certain forms of torrential weather will not happen more often than estimated.
 

MODULATION AND REUSE SCHEMES

 At a given power receive level and modulation scheme, the fundamental capacity of all spectrum is the same, bit-for-bit and hertz-for-hertz. For instance, 64 QAM modulation provides 5 bits per hertz, QPSK modulation 1.6 bits per hertz, no matter what frequency band they are used in. However, in an effort to squeeze more distance out of system transmission specification, lower density modulation allows greater distance at a given power, but sacrifices data throughput rates. Thus, MMDS can make use of 64 QAM for its downstream links, giving a raw downstream capacity of about 1 Gbps for its 200 MHz of bandwidth. LMDS systems, however utilize QPSK, and therefore realize about 1.8 Gbps of raw capacity, even though they have five times the MMDS bandwidth. This allows LMDS systems to use a cell radius of 2 miles, rather than the much smaller radius that would be required for them to use 64 QAM.

The net effect of the small cell size forced on 24, 28, and 38 GHz systems is that they must repeat the spectrum often via many separate cells in the market. We will discuss the cost implications of this below. Since each cell repeats some portion of the spectrum in some form this multiplies the capacity. This is a fundamental spectrum reuse scheme, albeit one forced upon such high frequency systems by their propagation characteristics.
 

Figure 1:  30 Degree Sectorized Antenna Pattern.
Channel A = Red, Channel B = Green.  Channel Reuse factor = 6
 

Another form of capacity multiplication is to sectorize the transmission pattern. Thus, in one simple version, instead of transmitting channel A on a 360 degree pattern, and channel B on a 360 degree pattern (effective coverage of two channels), channels A and B would be alternated in pie shaped sectors. If, for instance, twelve thirty degree sectors were used, there would be six channel A's, and six channel B's, each capable of carrying completely separate data. Thus, the effective coverage would be 12 channels, or more if sectors were allowed to overlap. All omni-directional wireless data systems make use of this sort of method, either for outbound transmission, inbound reception, or both. (Figure 1)
 

CELLULARIZATION: OBLIGATION OR OPPORTUNITY?  High frequency systems (24 to 38 GHz) will transmit only short distances. This means that in order to cover large areas, these systems are obligated to cellularize. This obligation imposes costs that in turn require certain densities of customers to justify. MMDS systems will transmit 35 miles in a single cell site. Therefore, given flat terrain in either case, MMDS systems are not obligated to cellularize. It is very important to note that not being obligated to cellularize doesn't mean it won't happen. It means that the designers of MMDS systems can decide to cellularize where the economic opportunity lies. This can lead to much more economic deployment of MMDS.
 

Figure 2: A Three Cell Supercell Design

 In the case of the SpeedChoice's Phoenix deployment, a so-called "supercell" transmission design has been employed. This means that a major cell overlays the entire market, in this case broadcasting from South Mountain. The radius of this cell is effectively 35 miles. Return spectrum is sectorized into 45 degree sectors. A second cell site has been deployed to cover an area of terrain shadow generated by Shaw Butte. With this configuration, the system can obtain an estimated 85% coverage with two-way wireless HSA service. Additional cells can be deployed, but essentially only need be deployed to add capacity or fill in small gaps in line-of-sight. In other words, the business plan is secured with two or three cells, but is by no means restricted to two or three cells. The effective area covered by this deployment is 3,850 square miles. By comparison, an LMDS system built to cover the same area would require 306 cells, using a 2 mile radius. The graph below shows the number of cells required to fill a 3,850 square mile area as the cell radius varies.
 
 

 
 Bellcore, the engineering consulting firm formerly owned by AT&T, published a study and economic analysis of LMDS prior to the LMDS auctions. In it, the study assumed that in a market of 1.3 million homes, a 2.5 to 3 mile cell radius was used, and analyzed the economics of providing data, video, and telephony services. The conclusion was that only 25 cells covering only 2% of the land area should be built to yield an economic business. These 25 cells would each require capital expense, site acquisition, rooftop leases, fiber backhaul connections, etc. Clearly, the rest of the market area must be left uncovered to preserve favorable economics at this high band.
 

Figure 4:  Example: Coverage Comparison of 25 two-mile radius cells
and one 35 mile radius MMDS Supercell.
 

As a result of the system infrastructure economics demanded by the frequency propagation characteristics at these high bands, it is no surprise that Teligent, Winstar and ARTT have all stated that their current business plans are to serve businesses only . Considering that only certain high-end demographics would support HSA in residences, and further considering that those high-end demographic people tend to live in highly dispersed neighborhoods and housing, it would highly uneconomical to put up an expensive cell site covering a few square miles for what might at the end of the day be a few dozen customers.
 

BUSINESSES AND RESIDENCES: WHAT? IS THE DIFFERENCE?

 We all know what a business location is. It's a really big building where a lot of people work. We also know what a residential location is. It's a house where a few people live but don't work. And of course, the needs and nature of these two markets are entirely different. Not anymore . The fact is, the island of Manhattan aside, business people have been dispersing from the really big buildings for decades. During the 60's, 70's and 80's, the urban cores suffered meltdowns to rings of suburban-based enterprises. Businesses locating in the suburban ring 10 to 25 miles out enjoyed more space, lower cost of operations, easier recruiting, and greater flexibility. This "doughnutization" of the urban business profile is well documented, and followed the extensive development of the national highway system and the various municipal road systems that feed it. The next step began in the 90's and will carry over to the 00's.

This is the "atomization' of the workforce, the ultimate dispersal. Called by various names like telecommuting and the virtual office, it simply means that more and more people will still be gainfully employed working for large companies, but will be spending more and more time working at home. Of course, here it is not a road system that enables the transition, but a global information superhighway ??? the Internet. The Phoenix market is a particular example of the imperatives that drive telecommuting.

The City of Phoenix has a special telecommuting program in place that provides incentives for businesses to use it. In fact, there are financial penalties for businesses in Phoenix that do not implement telecommuting. On a smog alert day, large companies are required to keep at least 5% of their workforce home. As a result, certain large companies in Phoenix have over two thousand telecommuters each. There may well be enough telecommuters in Phoenix to populate a small city. As the Internet, through Virtual Private Networks (VPN), becomes the preferred means of most telecommuting strategies, companies more and more will wish to have umbrella coverage of an entire market by a single Internet provider. As usage gets more intense, it becomes extremely expensive to have an employee wait half an hour while a large spreadsheet or graphic downloads with a 56 Kbps modem. For only a few dollars more per month, the employee can save many hours per month, thus the economics highly favor High-Speed Access. The issue then becomes finding an HSA provider that covers the market.

Currently, wireless operators like SpeedChoice are able to do so instantaneously, facilitating high-speed connections to dozens or thousands of sites. The SpeedChoice supercell design creates and umbrella that covers an entire market at once, enabling the connections of dispersed homes and businesses to a single high-speed network. On the other hand, high frequency networks like LMDS can only provide limited pockets of service and will not have ubiquitous residential coverage, if any.
 

CONCLUSION

 Certain technologies create niches no one knew were there. Kenneth Olsen, the CEO of the now defunct Digital Equipment Corporation, was quoted as saying "There is no reason for anyone to have a computer in their home." That's not as dumb a statement as it seems today. When he said it, there were no at home uses for a computer that seemed compelling. Such uses were developed once the platform was available. By merely existing, these new technologies create niches. Other technologies fill needs that are clear and well established. Certainly wireless competition to the RBOC monopoly falls into this category.

The best possible situation would be a combination of both. This would be a technology with a quick fix to an empty niche, but with enough new functionality that it will be used in unforeseen ways. HSA over MMDS should be in this category. The nature of Internet use is changing rapidly, and the need for Internet access is exploding.

Add to that the new technology's ability to reach both homes and businesses and get them up and running quickly and economically, and the possibilities are endless. This is true for SpeedChoice and all other vendors providing services using the MMDS spectrum.

---

SpeedChoice is headquartered in Shelton, Connecticut, and has exclusive rights to MMDS frequencies in major metropolitan areas of Arizona, Illinois, Indiana, Missouri, Michigan, New Mexico, Utah and Wisconsin More information on SpeedChoice's products and services is available at www.digitalchoicetv.net or http://www.speedchoice.com SpeedChoice is a wholly owned subsidiary of People's Choice TV Corp. (PCTV OTC: BB). PCTV has recently agreed to be acquired by Sprint Corporation, in a cash transaction that is projected to close during the summer.