Hi Inkyu, wow, that is *awesome* frequency accuracy!
You see a frequency offset range of 17 Hz in your example. Relative to 2.4 GHz (which I assume you're using as carrier frequency), that's 17 / 2.4e9, so approximately 7 ppb. Two unlinked oscillators that are not atomic clocks running this close together is so unlikely, you should be starting to play the lottery! So, my guess is that your IEEE802.11 receiver already does frequency correction, and you're observing residual frequency error after frequency correction, or that your two USRPs are sharing a 10 MHz reference. Is that possible? Also, how do you estimate single-digit Hertz accuracy with the amount of info in a WiFi preamble? Can you tell us how your estimator works? Best regards, Marcus On Mon, 2018-10-22 at 09:56 +0800, Inkyu Bang via USRP-users wrote: > Hi, all. > > (I recently shared this issue in USRP forum but it seems not properly > shared so I sent it again) > > I am testing frequency offset estimation by using the preamble of Wi- > Fi packet (IEEE 802.11). > > My setting is as follows: two USRPs are connected to one laptop (i7 > quad-core, SSD, 8GB ram). > > One is used for Tx. and the other for Rx. I chose 1MHz bandwidth and > set proper packet length and inter-packet time (50ms) so packets are > almost continuously transmitted. > > During the test (very stationary environment), 1.5 m distance between > Tx. and Rx. and good SNR is guaranteed (by checking constellation). > > At the receiver side, I estimated frequency offset value of every > packet using the preamble of each packet. > > I compared these values. Of course, I know that the frequency offset > between two USRP varies due to several reasons (e.g., temperature). > > Anyway, we can see similar frequency offset values during a given > period (e.g., 5 sec) and this is what I exactly expected. For > example, 2001 Hz (packet 1), 2003 Hz (packet 2), 1997 Hz (packet 3) > and so on. > > The problem is that, sometimes, I see unusually a high frequency > offset value for some packets. Then, frequency offset values return > back to the reasonable value again. For example, 2001 Hz (packet 1), > 2003 Hz (packet 2), 2307 Hz (packet 3), 2310 Hz (packet 4), 1997 Hz > (packet 5) and so on. > > Can anyone help me to understand this phenomenon? > > Thank you in advance. > > - Inkyu > _______________________________________________ > 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
