Received signal strength in dBm.

[Ged Haywood]

Hi there,

On Wed, 7 Jul 2004 Daniel Faria wrote:

> [snip, snip]
> having a hard time getting signal strength in dBm for received packets.
> My setup: Keith Smith's linuxAP, w/ hostap-0.0.4 (modified by me).

Confucius said:

"Man with wrist-watch know what time it is; man with two wrist-watches
never really sure..."

Well he didn't really say that of course, but you'll get the drift. :)

It is not at all clear to me what you are doing when you make your
measurements, but it seems to me that you are not comparing like with
like.  You need to give much more detail.  You seem to think that
there must be something wrong because two of your numbers are rather
different.  What are your reasons for thinking that they should be the
same?  Are your AP and your wireless card using the same antennas and
cables?  Both for transmitting and receiving?  Have you experimented
(for example reverse the physical locations of the devices, or swap
the antennas and/or cables if it's possible) to see what happens?

I suggest that you make some sort of estimate of the numbers you are
expecting to see, and then work out why what you see is different from
what you expect (and it will be :), not why two unknown quantities are
different.  See

http://www.jubileegroup.co.uk/JOS/radio/powerbudget.html

for an example.

> When the AP is reporting -51dBm, the card is reporting -37dBm, with
> some random oscillation around both values.

You can always expect variations in signal strength measurements.  For
example you can make a 30dB change in your received signal strength by
moving an antenna just a few centimetres.  Interference, moving trees,
vehicles and people, passing storms etc. can all make big differences.

> My card uses txpower=15dBm, which is probably close to what the prism
> card in the AP uses [snip]
> The path loss in both ways should be similar.

When you're making measurements, 'probably' and 'should be' are not
acceptable.  A stated probability _might_ be acceptable under some
circumstances.  These are not the circumstances.

> So, it seems to me that dBmAdjust should be 115, or something close
> to that.  I could by no means find a reason for this +-15dB
> difference I'm getting.

Making measurements is an important and very much, er, underestimated
science.  You need to determine first of all what you are measuring,
what instruments you are using to measure it, what you believe are the
estimated precision and accuracy of the measuring instruments, why,
and under what conditions these estimates hold true.  Then you will
have to spend quite a bit of time convincing yourself that you can
make a reliable measurement.  Part of that process will necessarily
involve making changes (usually just one change at a time, to avoid
confusing yourself unduly:) which you think should have predictable
effects on the values which you measure, and then explaining to
yourself why the predictions and the observations differ, and why the
magnitudes, and perhaps other qualities of those differences are what
they are.

Then I'm afraid you will have to work out why you still don't fully
understand the results, but you'll be a little bit closer.

FWIW if you really do have -51dB received signal I don't think you
have much to worry about.  I routinely work with worse than -80dB.

73,
Ged.