Hello, I'm working on a TDOA based localization platform using 3 USRP X300's as receivers. I have them synchronized with a 10 MHz ref and PPS signal generated by an OctoClock. However, I'm having trouble getting reliable localization performance through this system. My TDOA measurements are not what I would expect for the geometry I'm using. If I have the USRPs flashed with the same FPGA image and I use the same sample rate (i.e. 200e6/22 = 9.0909 MHz) across them all, is it possible the hardware could still be introducing different delays through each receiver? Rich

On Thu, Jan 9, 2020 at 12:14 PM Richard Bell via USRP-users < usrp-users@lists.ettus.com> wrote: > Hello, > > I'm working on a TDOA based localization platform using 3 USRP X300's as > receivers. I have them synchronized with a 10 MHz ref and PPS signal > generated by an OctoClock. However, I'm having trouble getting reliable > localization performance through this system. My TDOA measurements are not > what I would expect for the geometry I'm using. > > If I have the USRPs flashed with the same FPGA image and I use the same > sample rate (i.e. 200e6/22 = 9.0909 MHz) across them all, is it possible > the hardware could still be introducing different delays through each > receiver? > Did you calibrate their phases using some type of calibration signal such that you know the relative phase offsets of each radio to each other? Do you need to not know this information? Brian

No I don't need to know phase information. I'm cross correlating the pairs of receivers and the location of the peak gives me the TDOA. If the hardware chains across different radios introduce different delays, that would invalidate the TDOA measurement. So long as the delay is the same through all the hardware chains, the TDOA estimate will be accurate. Can I assume the hardware delay through X300 USRPs with the same FPGA image and set to the same sampling frequency will be the same? On Thu, Jan 9, 2020 at 12:51 PM Brian Padalino <bpadalino@gmail.com> wrote: > On Thu, Jan 9, 2020 at 12:14 PM Richard Bell via USRP-users < > usrp-users@lists.ettus.com> wrote: > >> Hello, >> >> I'm working on a TDOA based localization platform using 3 USRP X300's as >> receivers. I have them synchronized with a 10 MHz ref and PPS signal >> generated by an OctoClock. However, I'm having trouble getting reliable >> localization performance through this system. My TDOA measurements are not >> what I would expect for the geometry I'm using. >> >> If I have the USRPs flashed with the same FPGA image and I use the same >> sample rate (i.e. 200e6/22 = 9.0909 MHz) across them all, is it possible >> the hardware could still be introducing different delays through each >> receiver? >> > > Did you calibrate their phases using some type of calibration signal such > that you know the relative phase offsets of each radio to each other? > > Do you need to not know this information? > > Brian >

On Thu, Jan 9, 2020 at 6:45 PM Richard Bell <richard.bell4@gmail.com> wrote: > No I don't need to know phase information. I'm cross correlating the pairs > of receivers and the location of the peak gives me the TDOA. If the > hardware chains across different radios introduce different delays, that > would invalidate the TDOA measurement. So long as the delay is the same > through all the hardware chains, the TDOA estimate will be accurate. Can I > assume the hardware delay through X300 USRPs with the same FPGA image and > set to the same sampling frequency will be the same? > I'd think the group delay should be pretty consistent - at least within 10's of nanoseconds of each other if the setup is identical. What type of variation are you seeing when you perform your cross correlations? How much variation are you able to handle? Brian >

Hi Richard, To clarify, are you using a common Octoclock, with the 3 X300's in the same location, or separate locations with 3x Octoclocks? Do you have equal length cables to the antennas and does the rest of the system match ? What is your RX frequency ? What daughterboard are you using? You may want to try using a decimation factor that will produce an non-fractional host sample rate, instead of 200e6/22 = 9090909.09090909_ MHz. Does running at 10 MS/s sample rate produce any difference in the result? Regards, Nate Temple On Thu, Jan 9, 2020 at 4:08 PM Brian Padalino via USRP-users < usrp-users@lists.ettus.com> wrote: > On Thu, Jan 9, 2020 at 6:45 PM Richard Bell <richard.bell4@gmail.com> > wrote: > >> No I don't need to know phase information. I'm cross correlating the >> pairs of receivers and the location of the peak gives me the TDOA. If the >> hardware chains across different radios introduce different delays, that >> would invalidate the TDOA measurement. So long as the delay is the same >> through all the hardware chains, the TDOA estimate will be accurate. Can I >> assume the hardware delay through X300 USRPs with the same FPGA image and >> set to the same sampling frequency will be the same? >> > > I'd think the group delay should be pretty consistent - at least within > 10's of nanoseconds of each other if the setup is identical. > > What type of variation are you seeing when you perform your cross > correlations? How much variation are you able to handle? > > Brian > >> _______________________________________________ > USRP-users mailing list > USRP-users@lists.ettus.com > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >

I meant to include this link with regards to the same rates in my previous email: https://files.ettus.com/manual_archive/v3.15.0.0/html/page_general.html#general_sampleratenotes On Fri, Jan 10, 2020 at 10:40 AM Nate Temple <nate.temple@ettus.com> wrote: > Hi Richard, > > To clarify, are you using a common Octoclock, with the 3 X300's in the > same location, or separate locations with 3x Octoclocks? Do you have equal > length cables to the antennas and does the rest of the system match ? > > What is your RX frequency ? > > What daughterboard are you using? > > You may want to try using a decimation factor that will produce an > non-fractional host sample rate, instead of 200e6/22 = 9090909.09090909_ > MHz. Does running at 10 MS/s sample rate produce any difference in the > result? > > > Regards, > Nate Temple > > On Thu, Jan 9, 2020 at 4:08 PM Brian Padalino via USRP-users < > usrp-users@lists.ettus.com> wrote: > >> On Thu, Jan 9, 2020 at 6:45 PM Richard Bell <richard.bell4@gmail.com> >> wrote: >> >>> No I don't need to know phase information. I'm cross correlating the >>> pairs of receivers and the location of the peak gives me the TDOA. If the >>> hardware chains across different radios introduce different delays, that >>> would invalidate the TDOA measurement. So long as the delay is the same >>> through all the hardware chains, the TDOA estimate will be accurate. Can I >>> assume the hardware delay through X300 USRPs with the same FPGA image and >>> set to the same sampling frequency will be the same? >>> >> >> I'd think the group delay should be pretty consistent - at least within >> 10's of nanoseconds of each other if the setup is identical. >> >> What type of variation are you seeing when you perform your cross >> correlations? How much variation are you able to handle? >> >> Brian >> >>> _______________________________________________ >> USRP-users mailing list >> USRP-users@lists.ettus.com >> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >> >