[ASRG] The mistaken axioms of wireless-network research

[John Sangster]

There was an enormous amount of research done in the early days of
cellular telephone development on radio propagation in an urban
environment.  And, while the quantitative results would change a lot
for today's wider bandwidth and different modulation schemes, it
would indeed still be true that coverage areas are not circular and
that there are many complex effects caused by multipath, Doppler
shift between receiver and transmitter due to relative motion, etc.
etc.

At microwave frequencies, even 900 MHz, it is common to account for
bending due to the atmosphere by assuming the Earth's radius is
about 1.333 times bigger than it is geometrically.  There is also a
diffraction effect, often modeled as "kife-edge diffraction" that
allows waves to bend slightly, with an attenuation dependent on the
amount of bending, when the waves pass over a hill, building top, or
whatever.  So even without fading and multipath, the basic coverage
situation is already more complicated than line-of-sight.

Multipath also does not merely cause attenuation of the overall
signal as if you turned down the transmitter power or inserted an
attenuator in the transmission line to the antenna.  Instead, it
distorts the instantaneous waveform at RF (i.e. the input to the
analog demodulator) and causes even more distortion after
demodulation.  The precise nature of that distortion depends on the
modulation method used, typically FM or PM.  That is, the ideal
demodulator (and there can be other effects due to a non-ideal one)
should reproduce at its output the instantaneous rate of change of
angle (FM demodulator) or the angle itself  (PM demodulator) of the
received wave.  (Typically, one or the other of these forms of
modulation / demodulation is present in a "digital" radio link
before you get to the digital demodulator that estimates the bit
stream.)  If that wave is the pure transmitted signal, what you get
is the undistorted analog  modulation waveform, which the digital
demodulator can then deal with easily.  If it's multipath-corrupted,
it's much more complicated, and the outcome depends strongly on the
relative strength and time delay of all the multipath components.
The analog demodulator is looking at the angle of a "phasor" as
comm-theory types call it (we laughed a lot when Star Trek adopted
that name for a weapon technology).  Normally for a pure RF carrier
that phasor merrily rotates around and around the origin in the
complex plane at a constant angular velocity.  For normal FM
modulation, the rate of rotation varies and the time derivative of
the angular velocity is used to represent the modulation.  (The
latter can be analog voice or a baseband signal representing a bit
sequence, or whatever).

Under certain conditions of multipath, the tip of that phasor can be
pushed in very close in to the origin, or even made to whip past it
on the "wrong" side, which results it very nasty spikes when you
look at the time derivative of the phase angle.  The bottom line is
that the data demodulator which now has to operate on this highly
distorted analog demodulator output wave will make lots of bit
errors under those conditions, lots more than you might preduct just
from the average attenuation of the RF signal at that moment.

Yes, it's complicated.  I think you would need a detailed model of
the demodulator just to get the right statistics, and probably a
Monte Carlo simulation or actual real-world field testing would be
the only way to get really accurate data on packet losses etc.

-John S.,
_______________________________________________
John H. Sangster  *  SPHINX Technologies, Inc.
Phone: 781-235-8800 / Direct Line: 781-235-8753
E-mail:  jhs at sphinxtek.com,  jhs at theworld.com



----- Original Message -----
From: "Bill Ricker" <wdr at world.std.com>
To: "Simson L. Garfinkel" <simsong at lcs.mit.edu>
Cc: <asrg at amsterdam.lcs.mit.edu>
Sent: Saturday, July 19, 2003 1:14 PM
Subject: Re: [ASRG] The mistaken axioms of wireless-network research


Thanks for the heads up on that one.

If you folks want to hear more about the falsehood of "line of
sight", I've
been studying VHF+ radio propagation  from both a
weather-forecasting and a
radio hobby point of view.

Bill
ars N1VUX


_______________________________________________
ASRG mailing list
ASRG at amsterdam.lcs.mit.edu
https://amsterdam.lcs.mit.edu/mailman/listinfo/asrg