Manav, I'll take a shot in the dark and point you toward this article I've referenced a couple times in the past:
https://witestlab.poly.edu/blog/why-does-my-received-spectrum-droop-at-the-edges/ In order to decimate data, the USRP will use halfband filters for decimation factors that are a multiple of 2, and will use a CIC filter for anything else. The use of this CIC filter causes rolloff similar to what you're describing. If you change your sample rate to be a decimation factor of 2 from the MCR and you see an improvement, then you've found the culprit. -Sam On Sat, May 23, 2020 at 3:08 PM Manav Kohli via USRP-users < usrp-users@lists.ettus.com> wrote: > Setup details: > USRP N210 w/SBX daugterboard > UHD 3.14 > GNU Radio 3.7 > Ubuntu 16.04 > gr-digital GNU Radio OFDM blocks used > > While attempting channel estimation for 64-subcarrier OFDM, I find that > for higher bandwidths, such as 20 MHz, the channel estimate shows the edge > subcarriers anywhere from 15-20dB below the central zero subcarrier, even > though the anticipated channel is relatively flat. > > Taking a larger FFT of the entire received OFDM packets shows the same > rounding as seen in the channel estimation. This rounding is roughly > symmetric, and occurs with any carrier frequency used. > > It was suggested to me that this may be caused by the antialiasing filter > on the SBX board, and any further help/advice would be greatly appreciated. > Is such rounding normal, and if so, how could it be compensated? > > Much thanks, > Manav > _______________________________________________ > USRP-users mailing list > USRP-users@lists.ettus.com > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >
_______________________________________________ USRP-users mailing list USRP-users@lists.ettus.com http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
