USRP: Time and phase synchronization for source localization ??

[ErikL]

Are you using the code from the example?

[USRP]

The code "AoA Detection 1^st TX Ref Signal.vi" (in MUSIC code) used a USRP REF TX. But I can't use it with RF generator.

Note:my principal programme is "niUSRP EX Rx Multiple Synchronized Inputs.v i " 

Best regards

[ErikL]

I believe the MuSIC example does what you need. It will require some work to study and extract the necessary pieces. From the example documentation:

6.3 Receiver:
Figure 6 presents the block diagram of AoA Detection RX Localization using MUSIC.vi. The RX VI sets up all the USRPs at the receiver to have the appropriate operating configurations. It first captures a specified number of samples per processing cycle. The RX VI then separates the calibration signal and the target signal using filters as specified by the user. The phase and gain calibration to eliminate the uncertainties is done based on the information provided by the calibration signal. After the target signal is calibrated, it is then fed to the MUSIC algorithm to detect the AoA.

[ErikL · Accepted solution]

To be clear phase alignment is a manual step done in post-processing based on the offsets and gains observed from the known reference signal.

[USRP]

THANK you for your help 

Now; the idea is clear 

Best regards

[tmy540]

hi Erikl:

          i am trying to do synchronization of four USRP 2932

          I' m sure that the GPS-Disciplined function of the USRP 2932 will perform the same behavior as the external rederence clock and pps signal

          because it can provide better frequency accuracy and UTC time

          But I am wondering that if i use the GPS-Disciplined function, will i still have the initial phase offset  between USRPs ?

          Thanks.

[baharehbb]

Hi,

I want to do a project mainly the same as yours. I would like to know if you could solve the phase offset problem with a RF tone generator and compensate the phase offset or not. 

Thank you,

Bahareh

[cochenob]

Thought I'd post this example program I wrote since this is an issue that always seems to pop up here (and it was something I struggled with initially as well).  Hopefully it helps others.

The main VI (Calibrate and Measure USRP phase.vi) is similar in function to the MUSIC example that often gets referred to here, but significantly scaled down and simplier to understand (plus it doesn't require you to have Matlab).

Basically it sets up two N210's with WBXs (or NI equivalent) in MIMO mode.  See the attached drawing for how to set up your experiment.  Similar to the MUSIC example...the "active" antenna measures the DUT signals.  An external source is fed into the other antenna channel, and taking advantage of the leakage in the switch, allows you to measure both the DUT tones and the "calibration" tones at the same time.  The program then allows you to choose the corner frequencies of the BPF's to separate each tone...calculate the relative phase of each....then subtract off the extra USRP phase from the measured DUT to get the actual phase.

A few caveats...make sure your "calibration" tone has a little extra signal level since it will get attenuated by 20-30 dB by the switch.  A noisy calibration tone may impact the accuracy of the final measurement.  I've included the mean and std dev of each phase calculation so you can get an idea of what the SNR is (in a roundabout way).

Also...I've noticed that even if the phase difference in the DUT is zero, I may still end up with a little bit of phase even after subtracting out the calibrated phase between USRPs.  I'm guessing this might be due to path length differences between the two antenna ports and going through the switch?  If I recall, it's on the order of a few cm's.  If you're doing relative measurements, then it shouldn't be an issue.  If not...then you'll have to calibrate that out as well.

Perhaps at some point I'll write up something more complete and post over on the Community page.  The MUSIC code is a good research example, but it's rather complex if all you want to do is measure a single DUT and wrap your head around what's going on with the hardware.

Hope this helps.

---

Brandon

[baharehbb]

Hi Brandon,

Thank you so much for your compelete and excelent explanation. 

Actually the project we want to do is finding direction of arrival with two simple antennas, so we need to measure the phase difference between the antennas.

We had USRP1 with two WBX and during the phase difference measurement we encounter to  this phase offset problem. 

Now we want to buy 2 NI USRP2920 and repeat the experiment. I have a few question:

first I don't know what is a DUT? It might be silly but I just start working in this area. 

Do you think this idea would work with USRP1. It is from Ettus and I work with GNURadio. Can I work with Labview with USRP1?

Also These N210 USRPs have 1 reciever and 1 transmitter, I can'f understand how it can recieve both calibration signal and the DUT signal? How can we measure the phase ambiguty from calubration signal since it is connected to the transmitter of USRP, and also how we can then apply the measured phase offset to our original signal?

Sorry for many questions, but since I don't have the device now and I can't see the VIs, I can't understand what is going on. My professor wanted me to see how to solve this issue and tell him what to buy. 🙂

Thanks,

Bahareh

[ErikL]

Hello -

Just making sure you had seen the following tutorial, example code, and video:

https://decibel.ni.com/content/docs/DOC-25716

Video:

https://www.youtube.com/watch?v=qBvpllCqDR0