Millimeter Microwave: Bandwidth at a Price

 
As we move past 10GHz, we enter what is known as the millimeter microwave region. In this
region, multipath ceases to be a problem because the radiated energy can be tightly focused
with a small, passive, dish-shaped antenna. A number of other constraints begin to manifest
themselves, however, as well as a couple of singular advantages. Generally, the shorter the wavelength, the more rapid the attenuation of the signal when it
is propagated through the air, and in the region above 10GHz attenuation rises sharply from an
initial level of 0.2 decibels (dB) per kilometer at 10GHz. To a certain degree, the ease of focusing
millimeter microwave signals into narrow beams has the opposite effect because of the intense
concentration of the RF energy within the beam, but still most network operators utilizing
these frequencies do not attempt to transmit more than a mile.
Transmissions also become increasingly subject to atmospheric conditions, particularly
rain. In fact, RF engineers have a term, rain fade, to describe the loss of transmission distance
during periods of heavy precipitation. Rain fade can be addressed by increasing transmitting
power, but in most places transmitting power is subject to regulation, and, in any case, RF
amplifier power tends to go down with increasing frequency because of the inability of the
power transistors used in the output stages of the transmitters to pass high-frequency waveforms
that are also high in voltage. Indeed, solid-state amplifiers capable of developing even
moderate power in the higher millimeter wave bands have been available only since the 1990s.
Thus, signal attenuation in open air over distance becomes a substantial problem.
Attenuation of the signal in the regions above 10GHz is attributable to two causes: water
vapor absorption and oxygen molecule absorption. Neither manifests a linear increase with
frequency, but instead both exhibit wild fluctuations, with peaks of absorption followed by
valleys and then further peaks, with an overall upward trend becoming evident. Incidentally,
the patterns for oxygen and water vapor absorption are quite different, and their peaks and
valleys do not coincide. Above 100GHz, oxygen molecule absorption quickly plunges to an
insignificant level while the water vapor absorption trend moves mercilessly upward while still
manifesting a series of high peaks and deep troughs as you go up in frequency.
From 10GHz to 30GHz, absorption of either sort is not a very serious problem, and only
one absorption peak of any significance is present, that occurring at 23GHz. Consequently, the
entire spectrum category is useful. Above 30GHz, water vapor absorption rises very steeply,
exceeding 10dB per kilometer at 60GHz. Notwithstanding, spectrum has been allocated for
broadband terrestrial use at 31GHz, 38GHz, and 39GHz, though water vapor attenuation is
already quite severe above 38GHz. Figure 3-1 shows how frequency relates to atmospheric
attenuation characteristics.
Another problem in the spectrum above 10GHz is the obstructing effect of not just walls
but of even light foliage. Transmissions in these bands need absolutely clear line of sight,
which obviously makes placement of base stations much more difficult.
A final problem, and it is a significant one, has been the equipment itself. The commercial
microwave industry has been making reliable if expensive equipment for use in the bands
below 25GHz for 30 years, but higher-frequency bands have been mainly used for radar until
quite recently. Only in the late 1980s did bands beyond 25GHz begin to be exploited for commercial
communications, the first use being in satellite systems where transmissions were
highly asymmetrical and the high cost of millimeter wave equipment satellite was not of much
significance in view of the already enormous cost of launch vehicles and the satellites themselves.
Early attempts to build terrestrial equipment for these bands were not very successful,
and only within the last year or so have millimeter wave data links reached a high degree of reliability.
Figure 3-2 shows an example of a widely used base station radio for millimeter wave
networks. Such equipment is typically more expensive than hardware intended for operation
below 10GHz. The cost of millimeter microwave equipment remains high today, though it has fallen a
bit, and thus the equipment lends itself only to use with high-value customers such as mobile
operators seeking high-speed backhaul and enterprises wanting medium-speed or high-speed
packet services. Nevertheless, the millimeter wave frequencies are not wholly problematic, and for a
certain customer base they are actually advantageous.
Generous spectral allocations are the norm in the higher bands, so the operator has more
inherent capacity with which to work. Now it is also true that bit-to-hertz ratios bear an inverse
relationship to frequency and are usually only unity in the millimeter wave bands, but with
bands spanning minimally several hundred megahertz and in some cases several gigahertz,
the loss in spectral efficiency is more than offset by the additional spectrum. Add to that the
fact that a millimeter wave airlink is generally much cheaper than a fiber link, and you begin to
see a business plan.
The higher frequencies lend themselves to narrow beam transmissions, which is also an
advantage in certain types of deployments. A tightly focused beam is ideal for a point-to-point
connection where the full spectral allocation is assigned to a single user because that same
spectrum can then be reused in another beam separated by only a few degrees with almost no
interference between the two. And because high-frequency signals are subject to rapid attenuation,
the spectrum can be reused in an adjacent cell whose center is as little as couple of
kilometers away.