[MLB-WIRELESS] 2.4GHz through obstructions (WAS: Colorbond/roof insulation interference.)

John Dalton john.dalton at bigfoot.com
Tue Feb 18 12:42:06 EST 2003


I realise the author has qualified these numbers with
"I wouldn't rely on it entirely", but I would have to
go further and say I would be very cautious about relying
on these numbers at all.  No slight is intended to the
author, especially as he has already acknowledged the
fallibility of the numbers.  Please take this as
constructive criticism.

As an example, the attenuation of a metal roof is quoted to
be 12dB (ie. 1/16th of the power is transmitted through the roof).
In reality attenuation through a conductive media, such as metal, is governed
by an approximation called 'skin depth'.  This is the depth you have
to go into the material for the field strength to drop by 37% (=1/e).
Looking up a text book such as Frankl (page 268), the equation for
skin depth is:
    (skin depth) = sqrt(2/(pi.f.u.g))
where: pi = 3.141... (ratio of circle circumference to diameter)
       f = the frequency of the radiation (=2.4GHz for WLAN)
       u = permeability (approx 0.0001 N/A^2 for steel)
       g = conductivity (approx 10000000 /(ohm.m) for steel) 
So for steel at WLAN frequencies, the skin depth is around 0.5um
(1um = 1 millionth of a metre).  For a 1mm sheet of roofing steel
the portion of the field which makes it though is around 2.72^-2000
which means the attenuation due to a roofing sheet is of the order
of thousands of dB (not 12dB).  In plain English this translates
to: iron roofing does not let radio waves though.

In practice, attenuation depends on the composition of an
obstruction as well as it's mechanical construction, in
particular its thickness.  For objects of finite extent
width and height come into play.  Also cavities must
be accounted for.  As you can see, it is quite difficult
to make reproducible attenuation measurements for complex objects.

I suspect that the numbers in the table were measured by putting
a WLAN transmitter and receiver on opposite sides of the object
in situ, looking at the received signal strength then guessing
what the signal would have been without the object in place.
This is not a reproducible technique and the measurements are highly
specific to that set up.

For example, in the case of the steel roof, assuming the 12dB was
actually measured, the signal would have to have been due to a signal
traveling through the ceiling (or any gaps in the roof) and reflecting
in a complicated way off the contents of the house.  Any similarity to
your own situation will be by chance.

The 'proper' way to measure transmission is to take a flat
sheet of the material to be measured.  This sheet should be
large enough that there will be little propagation around it
and it is effectively infinite.  Put a receiver and transmitter
in an anechoic chamber (a room with absorbing walls) and measure
the received signal.  Drop your sheet of material into the radio
beam and measure the received signal again.  The difference is
the attenuation.  Also you should do your measurement with different
thickness sheets.  Waves from the top and bottom will interfere,
so you the transmitted power will not simply reduce as thickness is
increased.  Rather you will see a sinusoidal variation.  (Ideally
results should be expressed as absorption, reflection and transmission
coefficients, which requires reflected power to be measured as well.)

Not everyone has an anechoic chamber, but I gather a football
field makes a pretty good approximation since it has nothing on it
to reflect signal (haven't checked this by experiment though)

Applying these results to the real world, requires a complicated
simulation taking lots of measured reflection, absorption and transmission
coefficients into account.  Alternatively with enough experience one can
make an okay guess using rules such as "metal reflects radio waves",
"dielectrics attenuate radio waves", "waves interfere" and so on.

It is possible to construct simple tables as posted by Simon, but they require
averaging over huge sets of measurements and should only be interpreted
in a statistical sense (and should include details of sample size,
sample spread and measurement technique).

In general, it is good not to trust any measurements unless a reliable
reference is given, or a reproducible measurement technique
is documented.

This mail is probably a bit over the top but I had to debunk these numbers
before they attain 'urban myth' status and throw future pathloss
calculations out of kilter (the numbers have already
been posted to Sydney Wireless).

Best wishes
John


PS.  Note that the 'skin depth approximation' relies on the media being
conductive.  Don't go blindly applying it to every situation.



To unsubscribe: send mail to majordomo at wireless.org.au
with "unsubscribe melbwireless" in the body of the message



More information about the Melbwireless mailing list