Re: [Discuss-gnuradio] OFDM channel estimation

[Martin Braun]

On 10/04/2016 06:25 AM, Andreas Weinand wrote:
> Hello,
> 
> i'm currently dealing with OFDM channel estimation and try to read out
> the estimated values for this purpose. It works when i take the ones
> that are given by the tag debug block (Key: ofdm_sync_chan_taps). But
> somehow the values look a bit strange to me. Can anybody explain why
> there are some carriers in the upper frequency range that are not
> allocated, although they should be (see the plotted values). I did not
> change the number of active subcarrier which is 48. So carriers should
> be allocated up to carrier 59.
> I have used N210 USRPs und SBX daughterboards to get these values, but
> the same thing happens when i run the ofdm receiver in a simulation. The
> magnitude of these upper carriers is still approx. 1 Decade below the
> magnitude of the other active carriers in that case like in the OTA
> setup. Can anybody help with this problem?

Andreas,

as you can see, the length of the channel estimate vectors is equal to
the FFT length. However, channel estimates are only calculated for those
bins that are actually occupied. Here's the code:

https://github.com/gnuradio/gnuradio/blob/807bba681d56589d2f02cd6227181d9abc796e17/gr-digital/lib/ofdm_chanest_vcvc_impl.cc#L198-L210

There's several reasons why the length of channel taps is FFT length
instead of the number of carriers, but one major one is that we shift
the channel taps around according to the coarse frequency offset.

Cheers,
Martin