Re: [Discuss-gnuradio] OFDM problem with USRP and GRC

[Marcus D. Leech]

On 04/25/2011 09:15 PM, killol wrote:
> Hello all
> I am also trying to get OFDM mod and demod blocks already present in
> gnuradio distribution to work.
> The blocks work when used on the same computer. Once I try to receive on the
> other USRP, the OFDM demodulator does not demodulate any signal. 
> Can anyone please let me know what problem might be there.
>
> Someone suggested that there might be synchronaztion problem because of
> which the receiver frequency drifts and does not demodulate the transmitted
> frequency. 
>
> Can someone please help me regarding the same.
>
> Best,
> Killol
>   
>   
Minor frequency errors are an inevitable part of any radio channel.  The
impact of these errors
  is more or less severe, depending on the modulation technique in use.

PLL synthesizers are at the mercy of the crystals that are used to
provide a reference clock for them.
  The crystal oscillators on the USRP family aren't top-of-the-line, but
they're quite good--roughly
  +/-20PPM or so.  So, what that means is that for every MHz of center
frequency, the actual
  frequency could be
  off by +/- 20Hz.  For every 1GHz of center frequency, it could be off
by as much as +/- 20kHz.

For wideband conventional modulation schemes, an error in center
frequency of up to +/-20kHz
  won't make that much difference, because the error frequency is much
less than the total bandwidth.

But in OFDM, where there a plethora of narrowband frequency bins,
offsets can cause real problems.
  So your receiver has to try to frequency track the transmitter, or you
need to use an external
  reference clock, which on the USRP1, you can't do (unless you use a
3rd-party device like the
  clocktamer board).  I'm not an expert on OFDM, but isn't the existence
of a "pilot channel" there
  to help frequency-track the receiver?

-- 
Principal Investigator
Shirleys Bay Radio Astronomy Consortium
http://www.sbrac.org