Hi AB, Let me make a guess here. Because of the clock is not perfectly synchronized, there is non-linearity. Therefore the channel estimator gives multiple coefficient. Try connecting 2 USRP with the same reference clock and pulse per second signal and see if there are still multipath.
Regards, Edwin Bakshi, Arjun <bakshi...@buckeyemail.osu.edu>于2017年8月31日周四 下午3:07写道： > Correction: The thing I mentioned about all taps being phase shifted by > the same amount isn't always observed. > > > I tried it again with the same USRP, (and 2 USRPs with a MIMO) and I still > get different channels every time I start the flow graph. > > > Identifying multiple taps: the channel estimate I get has >1 coefficient > that is significantly greater than a noise like value. Figure attached. > Shows only abs(chan). > > > Thank you, > > > AB > ------------------------------ > *From:* Edwin Li <lbyhp...@gmail.com> > *Sent:* Wednesday, August 30, 2017 11:49:30 PM > *To:* Bakshi, Arjun; usrp-users@lists.ettus.com > *Subject:* Re: [USRP-users] Initial phase offset at TX/ DC offset and LTE > > Hi AB, > > I suppose you are using 2 USRPs, right? Each USRP has its own clock. These > clocks have different phases. That's why you see a phase offset. > > As for the multi-tap channel, how do you identify it? > > Not sure how LTE eliminates the phase offset. But in my communication > program, I use preamble for phase estimation, then use costas loop to > eliminate the offset. > > Hope it helps you. > > Regards, > Edwin > > Bakshi, Arjun via USRP-users <usrp-users@lists.ettus.com>于2017年8月30日周三 > 下午9:09写道： > >> Hello everyone, >> >> >> I have a setup that records the channel between a tx and rx while they >> are connected by a wire. However I've noticed that while the channel is >> constant over one run, when I restart the code/flow graph I get a different >> (but still constant) channel. >> >> >> Can anyone shed some light on why this is happening? >> >> >> Also, I'm observing a multi-tap channel even on a wired connection. That >> seems suspicious to me. I've however verified that my estimation code is >> correct based on simulations. I'm using an N210 with a WBX board. >> >> >> Lastly, does LTE do anything to force the same offset or eliminate random >> offsets at the transmitting node? >> >> >> I searched the srsLTE github repo and found some variable >> tx_corr_dc_phase/gain, that suggests a fix, but I'd like some more >> inputs on this. >> >> >> github link: >> >> >> https://github.com/srsLTE/srsLTE/blob/590fff9c69214b4858de2e987410f0d24ad4275c/srsue/ue.conf.example >> >> Thank you, >> >> >> AB >> _______________________________________________ >> 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
