By Jeffrey Krauss, intellectual property developer and President of Telecommunications and Technology Policy

The auction prices will depend on the marketplace opportunity for the service that the spectrum will be used for, but also on the effect of rain.

Upcoming microwave auctions

The major auction in 1997 or early 1998 will be Local Multipoint Distribution Service (LMDS), for a license of 1150 MHz of spectrum at 28 GHz in each of about 500 Rand McNally Basic Trading Areas (BTAs). These will be used for point-to-point and point-to-multipoint communications networks, to carry two-way voice, data and entertainment video programming. By comparison, the earlier auctions were for licenses of only 30 MHz of spectrum, but at a much lower frequency range, around 2 GHz.

Radio propagation

There are a number of factors that affect radio propagation—the distances that radio signals will travel. At frequencies above about 1 GHz, radio signals travel in straight lines and do not easily bend around obstacles, so path blockage is the most important factor. Line-of-sight paths can be blocked by terrain, by foliage and by buildings. Other propagation factors include attenuation caused by rain drops, fog or dust in the atmosphere, absorption due to atmospheric gases (oxygen and water vapor), interference caused by multipath signal reflections, and a variety of other factors. These are described in more detail in ITU-R Recommendation PN.530-5 and in radio communications textbooks.

While many of these factors are always present, rain occurs infrequently, and the heavier the rainstorm, the less frequently it occurs. So the attenuating effect of rain is treated statistically. In engineering a radio link, you decide what percentage of the time you can live with a link outage caused by rain, and then look up in a table the rain intensity in millimeters per hour corresponding to that percentage.

For example, if you demand an availability of 99.99 percent, so that the link outage is less than 0.01 percent of the time, or 52 minutes per year, and you are located in the northeastern part of the U.S., you find that you must design your radio link to overcome a rain intensity of up to 42 mm per hour. But if you can accept only 99.9 percent availability (525 minutes of outage per year), then you need only design your radio link to overcome a rain intensity of 12 mm per hour.

As the frequency increases, the rain attenuation becomes worse. For 99.99 percent availability on a 5 km path length at 10 GHz, the rain typically produces an attenuation of 5.5 dB. At 18 GHz, the attenuation increases to about 17.4 dB. At 25 GHz it is 29.6 dB, and at 40 GHz it is 54.2 dB.

For the same amount of spectrum in different frequency bands, this greater attenuation translates into lower spectrum value in several ways. First, rather than using 5 km path lengths, it becomes necessary to use shorter path lengths, only 2 or 3 km. This translates directly into higher system implementation costs, because it requires more transmitter sites to cover a metropolitan area. For example, a 5 km path at 18 GHz must overcome 17.4 dB of attenuation during rainstorms. But that same attenuation occurs on a 3 km path at 25 GHz. If transmitter sites must be spaced at 3 km rather than 5 km, it takes (5/3)2 = 25/9, or nearly three times the number of transmitter sites to cover an area.

Rain attenuation at higher frequencies can be overcome by using more spectrum, but using it less efficiently to preserve path lengths. A less efficient modulation method is more robust, but at the expense of decreased capacity. This is why, for example, over-the-air broadcast stations must use less efficient 8 VSB modulation, while cable TV can use 64 QAM or 256 QAM modulation and can derive a higher data rate. In addition, for example, 16 QAM modulation can carry a higher payload in bits/sec/Hz than QPSK, but it needs a much higher carrier-to-noise ratio (8 to 12 dB, depending on error coding methods).

The FCC looked at this issue recently when it relocated the Digital Electronic Message Service (DEMS) from 18 GHz to 24 GHz, and decided that DEMS needed typically four times as much spectrum at 24 GHz compared with 18 GHz in order to overcome the effect of rain attenuation, among other factors. Thus, on a dollars per megahertz basis, you could say that 18 GHz spectrum is worth four times as much as 24 GHz spectrum.

Spectrum value predictions

The broadband PCS auctions brought in about $20 billion for 120 MHz of spectrum. The LMDS auction will distribute 1150 MHz of bandwidth, or about 10 times as much spectrum. Will it bring in 10 times as much revenue? I think that $200 billion is far too much to expect. Spectrum auction values do not scale linearly with the amount of spectrum, and the difference in rain attenuation between 2 GHz and 28 GHz is enormous. But because of the amount of spectrum being auctioned, the LMDS service offers the opportunity to compete both with local telephone carriers and with cable TV. Companies that want to jump into these markets could place a very high value on the spectrum, in spite of rain attenuation. We'll see what happens at auction time.