Local Multipoint Distribution Service

A more aggressive strategy than MMDS is a new delivery service called Local Multipoint Distribution Services (LMDS), also known in Canada as Local Multipoint Communication Service (LMCS). LMDS is a two-way, high bit rate, wireless service under development by a variety of carriers to vastly increase bandwidth. The main differences between LMDS and MMDS are that LMDS occupies much higher frequencies (a disadvantage) but occupies much more spectrum.

Table 6-7. LMDS and MMDS Compared

	LMDS	MMDS
Frequency Range	28 GHz	2.5 GHz
Amount of Bandwidth	1150 MHz Block A 150 MHz Block B	198 MHz

Using just QPSK modulation, an LMDS Block A service provider can serve more than 2 GBps of traffic per transmitter site. With sectorization and possibly polarity diversity, individual users can get up to 100 Mbps of full-duplex data service, point to point. If an LMDS carrier had 1150 MHz of bandwidth, for example, it would be possible to use 500 MHz for broadcast TV, 50 MHz for local broadcast, 300 MHz for forward data services, and 300 MHz of upstream data. Using only the relatively robust QPSK modulation, this bandwidth can provide the following:

• All the broadcast channels of DBS (500 MHz)

• All local over-the-air channels (50 MHz)

• Up to 1 Gb of full-duplex data service (600 MHz)

In other words, the potential exists to offer more TV than satellite and more data than cable. This frequency plan is just one example of how a carrier could choose to offer service. Other carriers might choose to segment their frequencies differently and would be permitted to do so under FCC rules. If technological hurdles can be overcome and business issues can be addressed, LMDS offers the greatest two-way bit rate of any residential service, wired or wireless, at comparatively low infrastructure costs.

Background

In January 1991, the FCC granted a pioneer's preference license to CellularVision ( www.speedus.com , Nasdaq: SPDE), now known as speedus.com, to provide a one-way analog broadcast TV service in Brooklyn, New York. The preference license granted exclusive use of spectrum to more than 3.2 million households. The service used 1 GHz of bandwidth in the 27.5 to 28.5 GHz frequency band. Prior to CellularVision, this spectrum was normally for point-to-point use, but a waiver was granted so that speedus.com could provide a fixed cellular point-to-multipoint operation for video distribution (wireless cable). Cellularvision went public in February 1996 under the ticker symbol CVUS.

The rationale for granting the license was the claim that the company had overcome two problems associated with high-frequency wireless transmission. The first problem was how to get sufficient signal strength from high-frequency transmission. The company addressed this by transmitting in relatively small cells. MMDS can transmit 30 to 70 km in radius. LMDS covers 3 to 6 km, thus accommodating more and smaller cells.

The reliance on multiple small cells gave rise to the second problem. Small cells create problems of signal interference between adjacent cells. CellularVision attacked this problem by using polarity diversity. If adjacent cells use different polarity, then they can use the same frequencies without mutual interference. Armed with these innovations, Cellularvision obtained its Pioneer's license and became the first—and, to date, only—LMDS service provider.

CellularVision offered 49 channels of analog, one-way broadcast service. But analog TV was not a winning business when competing against Time Warner and other cable giants in New York City. So CellularVision, which proved the spectrum is usable but wasn't making any money, began to offer 48 Mbps Internet access instead and changed its name to Speedus.com. With the development of wireless digital technologies, renewed interest has been expressed in LMDS as a provider of two-way service by other companies in other cities seeking to offer Internet access. Accordingly, the FCC took action to provide more spectrum in the 28 GHz range.

The FCC issued rules governing the auctioning of bandwidth in May 1997. Auctions for U.S. spectrum began in February 1998 with 139 approved bidders. The rules (FCC Notice of Proposed Rulemaking [NPRM] 97-082) mandate the following frequencies to be used for LMDS service in the United States:

• 27.50 to 28.35 (850 MHz)

• 29.10 to 29.25 (150 MHz)

• 31.00 to 31.30 (300 MHz)

These figures represent a total available bandwidth of 1.3 GHz, which is more than twice the bandwidth of AM/FM radio, television, and cellular telephones combined.

The total of 1.3 GHz is offered in two blocks, called Auction Block A and Auction Block B. Auction Block A consists of 1.15 GHz and is located at the following frequencies:

• 27.50 to 28.35 (850 MHz)

• 29.10 to 29.25 (150 MHz)

• 31.075 to 31.225 (150 MHz)

Auction Block B is located at the following frequencies:

• 31.000 to 31.075 (75 MHz)

• 31.225 to 31.300 (75 MHz)

The segmentation of bandwidth creates greater separation between transmitting and receiving frequencies, eliminating the need for filters and ensuring efficient use of LMDS's spectrum.

The NPRM lists the major elements of the LMDS auction rules:

• The LMDS spectrum will be licensed by basic trading areas (BTAs), for a total of 984 authorizations and 1300 MHz of spectrum.

• Two licenses, one for 1150 MHz (Block A) and one for 150 MHz (Block B), will be awarded for each BTA. Note that a Block B spectrum has about the same capability as an MMDS spectrum holder, given that the MMDS spectrum holder uses a cellular infrastructure.

• All licensees will be permitted to disaggregate and partition their licenses.

• The number of licenses a given entity may acquire is not limited. Thus, a single spectrum holder could control Block A spectrum in a majority of U.S. cities. In fact, this occurred when NextLink purchased the spectrum of WNP.

• Incumbent local exchange carriers and cable companies may not obtain in-region 1150 MHz licenses for three years.

• LMDS may be provided on a common carrier or a non-common carrier basis, or both. Common carrier status means that the network provider would provide resale rights openly on a nondiscriminatory basis to content providers. Non-common carrier status means that the network provider would provide content itself and would not be required to resell the raw transport service.

• Licensees will be required to provide "substantial service" in their service areas within 10 years. This means that licensees cannot hoard the spectrum; the government wants it used.

• Incumbents in the 31 GHz band will be able to continue their operations but will receive protection from LMDS operations only in the outer 75 MHz (31.0 to 31.075 and 31.225 to 31.300) of the band. Incumbents will be given 75 days from the date of the publication of this item in the Federal Register to apply to modify their licenses to operate in the outer 75 MHz of the band, and an additional 18 months to implement those modifications.

• Bidding credits and installment payment plans will be available to small businesses and entities with average annual gross revenues of not more than $75 million.

The provision that prevents telephone companies and cable operators from obtaining in-region licenses for three years is based on the fear that if these groups were to win spectrum, they simply would sit on it for the 10 years allotted before they are required to provide substantial service. Taking the spectrum out of circulation in this way would reduce competition for the existing services. The historic reluctance of local exchange carriers to invade each other's territory, coupled with this provision, will have the effect of keeping them out of the auctions entirely.

The federal auctions began in February 1998 and went through more than 120 rounds of bidding over a period of two months. Eventually 864 licenses were granted, which raised a committed total amount of $578.6 million for the U.S. treasury. The three largest winners were these:

• WNP Communications, a group of venture capital funds (Norwest Capital, Chase Manhattan Venture Fund), which bid $187 million for 40 metropolitan-area licenses

• Nextband Communications, with major backing from Craig McCaw, which bid $135 million for 42 licenses

• Winstar LMDS, a subsidiary of Winstar Communications ( www.winstar.com , Nasdaq: WCII), which bid $43 million for 15 licenses

WNP, which paid $187 million for its spectrum, was acquired by Nextlink Communications www.nextlink.com , Nasdaq: NXLK), the parent of Nextband, for $542 million in January 1999. This transaction makes Nextlink the dominant LMDS spectrum holder, with more than 90 percent of the top 30 markets in the United States.

Unlike the local phone companies, long-distance carriers were not precluded from bidding. They seemed likely bidders because LMDS is a form of local loop bypass. Furthermore, there were no limits on the number of territories owned, so it was possible for a single long-distance carrier to have a seamless national broadband network. Nonetheless, long-distance carriers did not participate—perhaps there was the perception of a wireless bandwidth glut or a lack of confidence in the technology.

Furthermore, many metropolitan areas did not attract any bids at all, and the FCC will attempt to auction these properties at a later time. Other industry observers felt that $578 million was a relatively small amount of money raised in comparison with billions raised in the PCS auctions for narrowband digital telephone services earlier in the 1990s. These observations led some commentators to conclude that the auctions were not successful.

In any case, the relatively low amounts paid meant that the price of spectrum would not be a barrier to service rollout. Time will soon tell whether the auction winners, namely Nextband, got in at the right price.

Spectrum allocation varies in other parts of the world. The Canadian LMCS service is licensed at 27.350 to 28.350 GHz. In Japan, it is 25.25 to 27.50 GHz. In Europe, various bands between 24.5 through 29.5 GHz are planned (CEPT T/R 13-02). This should not present major problems because LMDS is not a mobile service, and hence roaming is not an issue.

LMDS Architecture

LMDS is a small cell technology, with each cell about 3 to 6 km in radius. Figure 6-4 shows a schematic of LMDS service.

Content acquisition at the LMDS head end functions similarly to MMDS, cable, and satellite. National television feeds are delivered by the programmer to a production facility. In many cases, these national feeds come from DBS, but the feeds also can come from other geosynchronous satellite transmission or high-speed wired services such as fiber-optic networks. Local content and advertising are acquired over the air, encoded into MPEG, and multiplexed with the national programming for local distribution. As in the case of MMDS, MPEG is an important facilitator of LMDS because it enables digital multiplexing.

Data services received from Web content providers are already in digital format but would need additional processing, such as encapsulation into MPEG and address resolution, before being transmitted.

The program mix is delivered by fiber to the LMDS broadcast tower. Generally, the LMDS head end and the LMDS broadcast tower are not co-located because the head-end production facilities would be shared among several towers.

An LMDS transmitter tower is erected in the neighborhood, and traffic is broadcast to consumers using QPSK modulation with FEC. It is possible to use QAM modulation, but QPSK was chosen because it is more robust than QAM 16 or QAM 64 and because bandwidth is so plentiful that spectral efficiency is not an issue.

Figure 6-5 depicts consumers receiving the signal on a small dish about the size of a DBS dish or a flat-plate antenna. The dish is mounted outside the home and is connected by cable to a set-top converter, much the same way in which DBS connections are made. The signal is demodulated and fed to a decoder. Unlike DBS, LMDS is capable of two-way service, so both TV sets and PCs must be connected to the satellite dish. Furthermore, a two-way home-networking capability must be supported instead of just the simple broadcast scheme of DBS. This places new requirements on home equipment, which will be explored in the next chapter on home networking.

In the return path, the customer transmits to the carrier using the same dish with QPSK modulation. A Media Access Control (MAC) protocol is required because the residences in the coverage area share the return spectrum. In Figure 6-4, for example, all three consumers in the middle LMDS broadcast boundary receive broadcast TV and Internet traffic on the same frequency, say, 27.5 to 28.35 GHz. Likewise, all consumers share the same frequencies for upstream transmission, say, 29.10 to 29.25 GHz. In this case, all three users would contend for upstream bandwidth.

A variety of proprietary approaches for the MAC protocol are under consideration. However, the ITU-T draft new recommendation for LMDS, called J.116 (J.isl), indicates that the MAC should be based on cable modem standards (Annex B of ITU-T Recommendation J.112). This is basically the DOCSIS protocol presented in Chapter 3, "Cable TV Networks." Thus, the cable return-path bandwidth arbitration protocol is broadening its scope to serve other Access Networks. This has the clear benefit of reducing component costs.

One way to increase the amount of frequency available is to use sectorized antennas. These antennas can broadcast to and receive from a pie-shaped segment of the footprint. If frequency hopping is used, it is possible to use different hop sequences with each sector. This technology is illustrated in Figure 6-6.

Jimmy, Rosie, Grandma, and Einstein are all transmitting on the same frequencies back to the head end. If no sectors existed, the four households could contend for common spectrum. However, sectorization enables the same frequencies to be reused in each pie-shaped area. This increases the amount of frequency available by a factor of 4, given 90-degree sectorization. Each sector can have different polarity. If frequency-hopping spread spectrum is used, each sector can have different hop sequences.

An LMDS tower and transmitter are projected to cost about $500,000 to $1 million each. CellularVision's original layout had 12 transmitters in place in Brooklyn and Queens, New York, providing coverage to 72 square km, which is about 1 transmitter per 6 square km. Assuming that each square kilometer contains 1000 residences, then 6000 homes would be passed per transmitter. This brings capital cost per home passed to $100 to $150. In less dense areas, the cost per home passed increases, but not necessarily linearly with respect to density. This is because it is possible to have fewer transmitters. Local conditions (such as rain and foliage) and the expected customer take rate all have a role in planning cell layout.

Challenges to LMDS

Despite the advantages of LMDS compared to wired networks (including large amounts of bandwidth and relatively low startup costs), the fact remains that only one LMDS provider exists today and that relatively few vendors are developing the product. Texas Instruments, an early LMDS provider with pilots in North and South America, sold its LMDS business to a European company, Bosch Telecom, which was in turn sold to a combination of Cisco Systems and Motorola, to form Spectrapoint. Likewise, Hewlett Packard sold its LMDS operation to Lucent ( www.lucent.com , NYSE: LU). These companies would not have sold these businesses had the challenges to LMDS been less daunting.

Competition

The FCC has begun the process of opening additional spectrum bands for fixed wireless applications, including 25 MHz at the 4.6 GHz level and 1.4 GHz at the 40 GHz level. Teligent (Nasdaq: TGNT, www.teligent.com ) operates its service at the 24 GHz spectrum tier. WinStar (Nasdaq: WCII, www.winstar.com ) and Advanced Radio (Nasdaq: ARTT, www.artelecom.com operate at 38 GHz. All three have point-to-multipoint services under way in multiple markets.

It is also possible to use newly licensed spectrum in the 2.3 GHz range. This range has the bland name of Wireless Communications Service, or WCS . WCS has an aggregate of 30 MHz available in four blocks. This is very low frequency with excellent propagation characteristics. It would seem that radios in the 2.4 GHz (unlicensed band) or 2.5 GHz (MMDS) ranges could be adapted to WCS. There are no service offerings in this space, but new startup operations are being funded by venture capitalists.

Finally, there is competition from cable, MMDS, and xDSL.

Signal Quality

At 30 GHz, the wavelength of the LMDS signal is about a millimeter in length, whereas MMDS wavelength is 11 times longer and DBS is longer still. With that small wavelength, LMDS signal quality is attenuated by raindrops or a drop of tree sap. LMDS also requires clear line of sight. Because of these problems, LMDS is usable only in relatively flat areas with relatively little foliage. It might be possible to overcome foliage problems with more efficient modulation and antennas, but doing so would incur higher transmitter costs and possible regulatory problems.

The whole issue of getting good signal integrity is still the subject of lively debate. However, field tests in the United States and Brazil (where it rains a lot) have given encouragement that these problems are on their way to being solved. In a Brazilian test run in 1996, Texas Instruments reported that equipment was capable of operating at a radius of 6.5 km. Still, doubts about signal strength persist on Wall Street, where LMDS is viewed primarily as a solution for flat, high-density, low-rainfall, low-foliage areas.

Cell Size

Smaller cell size increases infrastructure cost (necessitating more towers and transmitters) but improves coverage and reliability to consumers. The following list details the main engineering factors that can increase cell size:

• Increased transmission and receiver antenna height

• Increased power

• Overlapping cells versus single transmitter cells

Increasing the height of the transmission and reception antennas greatly improves reception by improving line of sight, mainly by transmitting over trees. The downside is aesthetics and cost. Operators will try to construct towers as high as possible, consistent with local ordinances and tower costs.

In rainy weather, the operator might increase transmission power slightly to get more energy through the network. No adaptive transmit power exists; instead, power levels are set during trials to provide the required coverage. This increases transmitter cost and can cause interference.

Another technique for better reception is to have overlapping cells. In a single transmitter cell, all homes must send and receive to a single tower. Hewlett-Packard found that a single transmitter would reach only slightly more than 60 percent of the homes in a cell due to line-of-sight problems. With overlapping cells, a given household could be within range of two or more towers and could select the tower that provides the best signal strength by measuring power levels. Studies by Hewlett-Packard have shown that overlapping cells can substantially increase penetration rate to nearly 100 percent.

The challenge for LMDS operators will be creating economical deployment strategies. As a point of contrast, the design of cell size and the location for LMDS is unlike that of cellular telephone. Cellular telephone is a wireless system often used by consumers while in transit. For such a system, it is important to start with large cells and gradually split them into smaller cells as usage grows to minimize handoffs. A handoff is the process of transferring the responsibility for a cellular voice call from one base station to another as the caller moves out of range of the original cell. During handoffs, voice calls can go silent as the cellular network sorts out which base station is responsible for maintaining the call.

No handoff problem exists in the LMDS case because none of the subscribers are in motion. This means that LMDS operators must start with small cells to maximize coverage and reliability. Such a process is costly and changes deployment strategies from the cellular model. Although broad coverage was the key to success in a cellular network, identifying and serving areas densely populated with potential customers is key to LMDS. Instead of segmenting big cells into small cells (such as with cellular operators), LMDS operators must start with small, well-placed cells and add more small cells as service levels increase.

Carrier Buildout Costs

Due to its small cells, LMDS requires many more transmission towers than MMDS. Because MMDS has about 10 times the transmission radius of LMDS, LMDS requires about 100 times more towers to cover the same area. In addition to construction costs, delays are associated with site selection, easements, permits, and negotiation of rentals. The result is that LMDS is significantly more expensive and time-consuming to roll out than MMDS.

Cost of Consumer Equipment

Unlike DBS antennas, LMDS dishes can transmit as well as receive. This increases cost for consumer equipment compared to DBS. Furthermore, installation costs are associated with LMDS; unlike DBS, for instance, an installer would be needed to align the antenna. Many analysts have concluded that DBS is the upper limit of what consumers will pay for digital TV equipment, and LMDS could push costs beyond that.

Standardization

LMDS system developers are proceeding on a variety of proprietary schemes. To address this problem, on March 11, 1999, IEEE chartered the 802.16 Working Group on Broadband Wireless Access (BWA, grouper.ieee.org/groups/802/16/). BWA "specifies the physical layer and media access control layer of the air interface of interoperable fixed point-to-multipoint broadband wireless access systems." The project authorization request will be applicable to fixed wireless systems operating between 10 and 66 GHz.

Standardization work has just begun and will need to conclude quickly to keep pace with the progress of competing Access Networks.