Re: [Discuss-gnuradio] Theory - GSM traffic detection with a USRP board

[Carlo E. Prelz]

First of all, many thanks for your kind answer.

        Subject: Re: [Discuss-gnuradio] Theory - GSM traffic detection with a 
USRP board
        Date: mar 29 ago 06 11:26:48 +0200

Quoting Martin Dvh (address@hidden):

> Nozema:
> http://www.nozema.nl/Content.asp?ID=6
> 
> Agentschap Telecom:
> http://www.at-ez.nl/dav/
> 
> FM and TV info site:
> http://www.radiowereld.nl/fmtv/

...
...

> For the GSM channels in use in Holland see:
> http://nl.wikipedia.org/wiki/GSM_(communicatie)#Frequentieverdeling

Very useful pointers. Especially the GSM channel allocation
information - so I know where I am to expect traffic when I test with
my KPN cellphone... As far as TV stations are concerned, I live on the
ground floor, well-protected from too many radio waves, it appears. 

> Probably your handscanner sees the GSM channels only as noise since
> the channels are much wider then a scanner can cope with (probably 25
> kHz max)
> The other possibility is that all traffic is going on in the high band 
> (1700/1800 Mhz)
> Or something is wrong in your setup.
> What antenna are you using?

All interesting observations. I have an old Yupiteru MVT7100 handheld
scanner - it was already old when I bought it second-hand back in
1999, but it was reputed to be of good quality. It uses its standard
telescopic antenna (not ideal for this, I believe). I have now set it
to wide-fm and it seems to pick up something more. 

> I hope this helped,

It surely has. I will dedicate one more coding day to this before
turning my attention myself to paid assignments for some time. But I
surely have received some encouragement from your comments.

I am setting up a utility of this kind: decimation is at 16, so the
read stuff covers 4MHz, corresponding to 20 channels. I set the
frequency, feed the URB, collect it, queue the data for processing and
repeat the loop 4MHz further on in the band. According to your
experience, should I use a higher level of decimation to reliably spot
the signals or is this resolution acceptable? I use packets of 2048*4
bytes (two daughterboards, I and Q data for each). And how should I
treat this 'bw' value? I expect it to be 'bandwidth,' but I found no
explanation apart from the fact that it must be between 1 million and
33 million. Setting it to high values results in somehow smoothing the
curve resulting from the discrete fourier transform, but not in
anything else from what I can see.

And another thing. I experience random hangs in data reception from
the USRP. I have seen this both with my software and with the gnuradio
fftw utility. The programs do not crash - simply, no data seems to
arrive from the USB port anymore. If I restart the application, all is
OK again. I am using firmware compiled from a recent gnuradio SVN
repository. Is this a well-known 'feature', for which there is some
known fix, or is it local to my place?

Again, many thanks

Carlo

-- 
  *         Se la Strada e la sua Virtu' non fossero state messe da parte,
* K * Carlo E. Prelz - address@hidden             che bisogno ci sarebbe
  *               di parlare tanto di amore e di rettitudine? (Chuang-Tzu)