Hi DingJing, I'm not sure what's going on here, but I think it could be likely that your estimate of the preamble-autocorrelation simply doesn't work all too well.
Generally: ditch benchmark_rx and the blocks it uses. They are really superseeded by the OFDM blocks that were introduced in 2013 – they are far better, and also, will simply give you the estimates you want, I think. Make sure you have a recent version of GNU Radio, and look for the rx_ofdm.grc example (likely somewhere in /usr/share/gnuradio/examples/digital/ofdm). The "Schmidl & Cox OFDM synch." block /actually/ has an output "freq_offset", and the "OFDM Channel Estimation" block adds a stream tag containing the full subcarrier offset (if any). ** Other than that: > I've been trying to figure out the frequency offset of USRP recently. Could you specify this a bit more? Best regards, Marcus On 19.05.2016 05:53, DingJing wrote: > Hi all,I've been trying to figure out the frequency offset of USRP > recently. So I run the OFDM benchmark program and output the > compensation value of fine offset.For detail,in benchmark.rx.py, we > import ofdm_receiver,and ofdm_reciver import ofdm_sync_pn to calculate > the fine offset and do time synchronization.The main procedure of > ofdm_sync_pn(using Schmidl and Cox algorithm) is as following: > > And I output the S/H(sample and hold) block to file sink block to > record the fine offset.However,the result figure are as following: > X label can be regarded as time,Y label is the output fine offset by > angle between -Pi to Pi.It is fiercely different from my expectation > in which the figure should be a straight line approximately.I wonder > why this happens.Thank you so much! > > > > > > > _______________________________________________ > Discuss-gnuradio mailing list > Discuss-gnuradio@gnu.org > https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
_______________________________________________ Discuss-gnuradio mailing list Discuss-gnuradio@gnu.org https://lists.gnu.org/mailman/listinfo/discuss-gnuradio