The year is 2002 and you live in the middle of nowhere. Entertainment is hard
to come by, but you鈥檙e not bothered鈥攖he delights of the electronic
superhighway recently rolled into your rural idyll, courtesy of a new
1000-channel television, Internet and telephone broadcast network. Eagerly, you
turn on your digital telecomputer to join in a live telecast of Star Wars
II鈥擳he Interactive Edition. Thunder and lightning rend the sky
outside, although you鈥檙e too busy bringing Darth Vader back from the dead to
notice. But just as you鈥檙e steering your zombie avatar into the Death Star to
wreak ghastly revenge, the screen suddenly goes blank. The cause of the problem?
Rain.
This scenario is not as far-fetched as the film. Television reception has
always been susceptible to problems caused by bad weather. Most viewers will be
familiar with ghosting, where the signals bounce off nearby buildings to create
multiple images and to 鈥渟now鈥� caused by a weak signal. But surely digital
television should be immune?
Alas, not quite. Some digital broadcast technologies planned for the next
century can be blocked by a whole new set of factors ranging from heavy rain and
fog to trees and buildings. There will be no ghosting or snow on digital TV
sets鈥攖he picture will either be near perfect or entirely blank.
Advertisement
The problem is the frequency at which the digital signals will be broadcast.
Today鈥檚 analogue transmissions use a frequency of about 600 megahertz and a
wavelength of about 50 centimetres and the first generation of digital
terrestrial broadcasts now being planned will also use this frequency. They
diffract over hills and round buildings so that reception is possible even if an
aerial does not have a direct view of the transmitter. This is how a single
high-power transmitter can cover a large area with a 鈥渇ootprint鈥�, or cell, some
50 kilometres in diameter. Of course, there are sometimes problems for receivers
behind large buildings or in deep valleys.
But the generation of terrestrial broadcasts beyond this will be transmitted
at much higher frequencies. In Europe, they will have a frequency of 42
gigahertz and a wavelength of only 7 millimetres and in the US at a frequency of
28 gigahertz and a wavelength of about 10 millimetres. The advantage of these
higher frequencies is that much more information can be crammed into the
signals.
But the physics of signal propagation is very different at these frequencies.
Millimetre waves do not spread out behind hills and buildings, so they cannot be
received unless the receiving aerial has a direct view of the transmitter. So
each digital transmitter will have a tiny footprint no more than a few
kilometres across.
Even in this small area the weather will play all kinds of tricks with the
signals. A heavy shower sweeping across the country could turn screens blank as
it moves through one cell after another. At other times, certain weather
conditions could carry the signals for many kilometres so that broadcasts from
one cell might interfere with others. Heavy rain is already a familiar cause of
poor satellite reception, which uses a frequency of 11 gigahertz.
Broadcasters accept that it will never be possible to get perfect reception.
The challenge is to work out how strong the effects of weather will be in
different parts of the world and how big the cells must be to provide the best
coverage at a reasonable cost. Broadcasters will even have to work out where to
put their transmitters so that each house in the cell will have a direct view of
them. These are not easy tasks.
The challenge of solving them has fallen to a team of European researchers
led by the Radio Communications Research Unit (RCRU) at the Rutherford Appleton
Laboratory in Oxfordshire. They are the first to admit that the physics of
signal propagation through the atmosphere is a complex science. The atmosphere
affects signals of different frequencies in different ways. For example, oxygen
molecules absorb signals at 60 gigahertz while water vapour absorbs them at 22
gigahertz, so signals at these frequencies are rapidly absorbed. This is one
reason why European broadcasters have chosen the 42 gigahertz 鈥渨indow鈥� which
lies between these absorption lines. Even in this window rain can have a
significant effect on signals because liquid water absorbs telecommunications
signals.
Absorption isn鈥檛 the only problem. The atmosphere also bends or refracts
signals by an amount that depends on its temperature. Layers of hot and cold air
can act like mirrors playing havoc with signals. Just how badly millimetre-wave
communications are affected by these factors is a crucial part of the work at
the RCRU.
Of course, nothing can be done to prevent these weather conditions, even by
researchers at the RCRU. Instead, their work focuses on understanding exactly
what effects these conditions have on the signals, how often the weather is bad
enough to prevent reception entirely and on ways of mitigating these effects.
They have made good progress on understanding the effect of reflective layers in
the atmosphere on millimetre waves.
鈥淯nder certain conditions of temperature and humidity, you can get large
changes in atmospheric refractive index over very small distances,鈥� says Ken
Craig, who heads the RCRU. Communications scientists have known for years that
signals can become trapped between these layers, a phenomenon known as ducting.
It can cause signals from one area to interfere with others many tens of
kilometres away.
Up, up and away
To work out how badly millimetre waves are affected by ducting, the RCRU team
must measure how strongly the waves are reflected in specific conditions. First,
they have to wait for the sharp changes in temperature and humidity that can
cause ducting to occur. Then they have to launch balloons carrying sensors into
these layers allowing researchers to work out how much has been reflected.
The work has thrown up some interesting results. It turns out that millimetre
waves are not as badly affected as longer wavelength signals. Turbulence in the
atmosphere makes the interface between layers ragged, at least on the scale of
millimetres. The result is that millimetre waves tend to be scattered out of the
duct rather than internally reflected.
Craig believes that ducting is worse at lower frequencies. On the one hand,
ducts tend to trap higher frequency signals more efficiently while on the other,
higher frequencies are more severely scattered. 鈥淪omewhere in the microwave or
ultrahigh frequency region, there is an optimum frequency where ducting is
strongest,鈥� he says.
Rain causes more severe problems. When radiation hits a raindrop, some of it
gets scattered and some of it gets absorbed, heating the drop by a tiny amount.
鈥淏ecause raindrops are typically a few millimetres in diameter, you get the
strongest interactions in the millimetre part of the spectrum,鈥� says John
Norbury, who recently retired as head of the RCRU.
It is possible to work out exactly how a raindrop of a given size and shape
will affect a signal of a given wavelength. 鈥淭he physical process is well
understood and the predictions are well-tested,鈥� says Craig. 鈥淭he uncertainties
come in the distribution of drop sizes and shapes in different types of
谤补颈苍.鈥�
A major part of the RCRU鈥檚 work is in building a database of these
distributions using rain radar measurements. 鈥淭emperate climates are
well-understood, but tropical climates less well so,鈥� says Craig. The unit has
recently installed a rain radar in Papua New Guinea and is planning another in
Singapore to improve the database. In addition, the team are mining data from
existing meteorological rain radar databases.
Armed with this information, the RCRU team have developed a computer program
that, by simulating local rain patterns, can work out how badly transmissions in
a given area will be affected by weather.
Simulated rain
This is a hugely important factor for broadcasters. The amount of time during
which a receiver gets good reception from a transmitter is known as the
availability of the service. Today鈥檚 satellite broadcasters aim for an
availability of about 99.7 per cent鈥攔oughly equivalent to service being
down for about one day a year. Future broadcasters are likely to opt for a
similar level of service. 鈥淒eciding the availability you expect of the service
determines the intensity of the rain that the system can withstand,鈥� says
Craig.
But this is only the first step. In addition, the RCRU must work out how the
local terrain and vegetation influences propagation. Trees are a big problem
since leaves significantly attenuate 42 gigahertz signals and wet leaves are
practically opaque.
But while trees are a nuisance, buildings can reflect signals into otherwise
poorly served areas. 鈥淪ome building surfaces can appear relatively smooth to 7.5
millimetres wavelengths,鈥� explains Craig. However, reflections can also cause
interference that ruins the signal.
To find out exactly what happens to signals, the RCRU has equipped a vehicle
with a millimetre-wave receiver mounted on a telescopic boom that can measure
the strength of the signals at various heights around obstacles such as trees
and buildings. The team has made extensive measurements in the streets and
country lanes around its fixed transmitter.
This data has allowed them to develop a second computer program that traces
the path of millimetre waves once they leave the transmitter. The process is
remarkably detailed and involves building a 3D computer model of the area in
which the transmitter will be placed.
The team has already started using stereo photographs from an aerial survey
of the nearby city of Oxford to create a computer model of the city. 鈥淔rom
stereo pairs, you can tell not only where the buildings are in a cell but how
high they are,鈥� says Craig, pointing to a computer model of the engineering
building at Oxford University which towers over the rooftops.
The program allows researchers to work out how many houses there are in a
given cell by the number of roofs that can be seen from above. The roof count is
important because this is where receivers are likely to be placed. 鈥淕iven that
your transmitter is on top of a tower somewhere, you can use the 3D tool to do a
visibility check. How many roofs can you see?鈥� asks Craig. It can then compare
various vantage points to work out which gives the best coverage. The virtual
model can even work out which houses will be blocked by trees or other
buildings.
The technology that will make millimetre-wave broadcasts possible is already
being tested all over Europe. The service is known as the multipoint video
distribution system (MVDS) and the trials are backed by the European Union and a
number of broadcasters, Internet service providers and transmitter
manufacturers. If the tests are successful, the technology could be rolled out
on a larger scale, using the RCRU鈥檚 software to determine exactly how the cells
should be laid out.
In this scenario MVDS will consist of a network of transmitters each covering
an elliptical area up to 5 kilometres across. The transmitters are low power
devices鈥�0.5 watts compared to kilowatt transmitters used today鈥攁nd
the receivers will be small horns about 15 centimetres in diameter with an angle
of view of only 2 degrees. This should prevent interference from neighbouring
cells using the same frequency.
The technology has to compete with other methods of transmission. For
example, cable companies already have a strong foothold in many cities. However,
MVDS will be a much cheaper way of providing cable-like services to rural
communities where the cost of laying cable is high. Just how popular it will
become remains to be seen.