Martin,

I think my question would have been clearer if i had said that when I

talk of symbol rate I do not define a symbol to be the OFDM symbol of

length

fft_length+CP_length, but its fft_length+CP_length components.

So for me symbol rate is

symbol_rate = (fft_length+CP_length) x OFDM_symbol_rate

The grc examples are defining this as the sample_rate, which

implies that they use one sample for each symbol (my definition).

In any case, from your answer I understand that the S&C operates on

ONE sample per symbol (my definition).

I still have trouble getting anything out of the S&C synchronizer:

The frequency offset output responds as expected when i introduce a

freq error but the “detect” output does not give anything out

even with noiseless input…

I have uploaded the follwoing grc files

http://web.eecs.umich.edu/~anastas/gnuradio/tx_ofdm2.grc

http://web.eecs.umich.edu/~anastas/gnuradio/rx_ofdm2.grc

which, illustrate my point. Any thoughts?

thanks for the help,

Achilleas

On Sun, May 19, 2013 at 11:45:42PM -0400, Achilleas A.

wrote:

I am having trouble setting some parameters and getting anything out of it…

In particular

in ofdm_rx

- is the processing done on an 1 sample per symbol basis at the Scmidl&Cox ?

I don’t understand the question. The correlator has no info about OFDM

symbol boundaries.

In that case the input sample rate has to be set to the symbol rate?

(what is called samp_rate in ofdm_tx)

This is the input sample rate. The bit rate depends on the number of

carriers used, the length of the CP and the modulation type on the

payload. The OFDM symbol rate is the sample rate / (fft_len + cp_len).