The Challenges of Wireline and Wireless Delivery

We tend to think of copper access as being bandwidth rich, but in practice we become
frequency-limited. Attenuation increases rapidly with frequency, which means the distance
over which we can travel becomes limited as frequency increases. In practical and
economic terms, it is hard to access frequency above 1 GHz. It is, however, a consistent
delivery medium and can be very adaptive in terms of bit rate (the whole idea of ADSL).
RF wireless access is not bandwidth-limited. There is plenty of bandwidth available
(270 GHz at least), but we are short of RF power. We can increase RF power by increasing
network density. Although this has a cost implication, it does provide us with
almost infinite bandwidth. However, RF is an inconsistent delivery medium that we
need to tame by using measures like adaptive power control. However, there are times
when a user will just simply be out of radio range. Depending on network build-out,
there will always be coverage black spots both in urban and rural areas. Typically
between 10 percent and 20 percent of a developed country with a mature network
build-out will still have either marginal or nonexistent coverage. It would be implausibly
impractical and expensive to provide the 99.999 percent availability delivered by a
wireline network.
Most wireline networks were built in the era of national telecom monopolies where, in
return for the right to a monopoly, the telco had a statutory obligation to provide service
to all subscribers. Many wireline networks have been amortized over many decades, a
luxury not available to wireless network owners. Although wireless theoretically offers
coverage cost benefits when compared to new build wireline, these benefits disappear
when compared to a legacy wireline network fully amortized many years ago.
In a wireless network, it is still expensive to get enough RF power distributed
widely enough to give good consistent coverage. The RF power requirement is dictated
by the quality requirement. Wireless network density is normally planned so that
even in coverage areas, the typical bit error rate will be 1 in 10-3, rather than the 1 in 1010
available over copper or 1 in 1012 available in the optical layer.
Infrared works in between 850 and 950 nanometers (100 to 118 Terahertz)�"
potentially a bandwidth of 18000 GHz, although typical bit rates are a power-constrained
2 to 4 Mbps. Infrared is also, like RF, an inconsistent and implicitly discontinuous delivery
medium, working best with a clear line of site between sender and receiver. As an example,
the ETSI/ARIB IrDA AIR (Area Infrared) specification delivers up to 4 Mbps over 4
meters or 260 Mbps over 8 meters with a 120° beamwidth.
As mentioned earlier, optical access promises almost infinite bandwidth. Even just
taking C Band and L Band (80 nanometers) gives us potentially 10,000 GHz of bandwidth
(10 Terahertz), and the optical layer is, of course, a consistent delivery medium.
However, optical fiber is not much use for mobility users. Free space optical transport
shares many of the drawbacks of the radio physical layer�"high attenuation
(including fog and rain dispersion) and the need (in common with higher RF frequencies)
for line-of-site communication.
We can never match the consistency available from wireline copper or optical fiber.
RF or optical free space transmission is inherently inconsistent. We can, however, manage
consistency as a quality metric in the same way we can manage bit error rate, delay,
and delay variability.347