I am assuming my last message may have slipped through. Trying one more time.. ________________________________ �����: Nik B. <nikkei16@outlook.com> �����Օr: 2016��8��28�� 5:53 ����: Marcus M��ller; Nik B. ����: RE: [USRP-users] Fwd: Re: Changes in carrier phase due to Doppler Hi Marcus, I was wondering if you have had time to read this mail... N. ________________________________ �����: USRP-users <usrp-users-bounces@lists.ettus.com> �� Nik B. via USRP-users <usrp-users@lists.ettus.com> �δ��������� �����Օr: 2016��8��19�� 18:38 ����: Marcus M��ller; usrp-users@lists.ettus.com ����: Re: [USRP-users] Fwd: Re: Changes in carrier phase due to Doppler Hi Marcus, I am trying to understand some of the things you wrote and also some from the patent. (1) .. at 920 MHz, you'd see 36.8 full oscillations...how did you get this no. (36.8)? (2) .. your [phi] knowledge woud be totally worthless ...why? (3) .. relative phase measurements obviously have a wavelength ambiguity... are you stating the general truth here, that is, in a "round-trip" (Tx1 -> Tx2 ->Tx1), it's a 2pi = wavelength ambiguity, right? The patent author mentions one half wavelength (item no. 0087 in the patent pdf version, please get from here:http://www.freepatentsonline.com/y2015/0346332.html) .. I wanted to make sure that you (and the author) are talking about the same thing. Let me quote a few lines from the patent: //----------item #0087-------- Carrier phase measurements typically have a measurement domain of a full wavelength, or 2�� radians, on each leg of a round-trip measurement. Some CPR measurements may have a measurement domain of one-half wavelength, or �� radians, when they are determined by a round-trip measurement. These measurements are said to ��wrap�� at a CPR measurement ambiguity, such as, for example, equal to one-half wavelength (��/2). //--------------------------------- (4) .. on each leg.. shouldn't it be just pi radian, instead of 2pi? Then in the second sentence, the author talks about "measurement domain of one-half wavelength, or �� radians" ... "when they are determined by a round-trip measurement". I am struggling to understand if the definition of "round trip" has changed. I wanted to know, these things, I mean the ambiguities .. do they change all the times (measurement by measurement)? I mean, could some measurement be wrapped ��/2, while some others be 2 * (��/2), yet some other be 3 * (��/2), etc.? I have just quoted above without all the contexts, but I hope you remember why you said that. Thank you for your time and insight, N ________________________________ �����: USRP-users <usrp-users-bounces@lists.ettus.com> �� Marcus M��ller via USRP-users <usrp-users@lists.ettus.com> �δ��������� �����Օr: 2016��8��16�� 21:13 ����: usrp-users@lists.ettus.com ����: Re: [USRP-users] Fwd: Re: Changes in carrier phase due to Doppler Hi Andy, I was guessing that two different frequencies were used in the patent for at least 2 reasons: a. it was an FDM communications system between interrogators and transponders b. two separate frequencies might be used to remove Doppler ambiguities Yep, b. was my central assumption / motivation, too. (also, claim 2) At any rate, things are getting quite involved here, and I think I'd really have to read through 32 pages of patent... might not really work out overly soon :) Looking at this from a higher level, and taking a bit of the math back out of it: * relative phase measurements obviously have a wavelength ambiguity * by using multiple frequencies and combining the knowledge gained, that ambiguity might be reduced/eliminated * looking at this as a information system, every "trip" between TX1 and TX2 has the potential to reduce uncertainty about the distance and relative speeds * this is a system that combines single-ended estimates by communicating them * the ultimative information sources in effect are - phase and frequency of the other station as observed using one's own phase and frequency, done by both, - knowledge of a physical model of reality, - and stateful observation, meaning that from previous measurements a change in phase might be extrapolated, representing a relative movement, predicted from observed Doppler * this demands "good" local oscillators, meaning that oscillator offset and drift need to be determinably for a time frame sufficiently large for observation to allow "calibration". Best regards, Marcus On 16.08.2016 13:45, Andy Walls via USRP-users wrote: >Hey Andy, >> 1. a frequency estimation/tracking and correction block that outputs >> the frequency correction between transmitter and receiver >> (pll_freq_det maybe?) and also performs the correction >> (pll_carriertracking maybe?). This correction is the doppler >> frequency plus the frequency offset between the two receivers. >> >> 2. Assuming the frequency offset between the transmitter and receiver >> is relatively constant across some number of measurements, it can be >> estimated and subtracted out. >Yep, essentially, higher-level "knowledge"; Doppler necessary changes >over time, clock offset shouldn't that much. But then again, let's >assume "not driven by sustained explosion at rear end of TX2" speeds >(i.e. automotive) of maybe $v=30\frac{\text m}{\text s}$, then the >relative Doppler shift would be $1+\frac vc \approx 1+ \frac{3\cdot >10^1}{3\cdot 10^8}$, so the frequency error of a Autobahn-speed car >would be around 100 ppb due to doppler �C so yes, that would be >significantly above drift of a working GPSDO, but not above what you'd >see e.g. with a TCXO. Yeah, I'm used to working with higher speeds. I tend to think of doppler as being a measurable frequency shift, not swamped by system noise. For low speeds, yes, this is likely not be the case. >> 3. Assuming the measurement message has some well known preamble, and >> that the received signal has been frequency corrected (in 1. above), >> then the phase of the peak of the complex cross correlation of the >> signal preamble with a reference preamble should be the carrier >> phase. This is what the corr_est block outputs. This phase will >> include OTA propagation as well as delays through coaxial cables and >> other system components, etc. Those will need to be calibrated out. >Yes! But: wouldn't that imply we know the phase of the TX LO, which >would imply we needed to track the LO (PLL tracking?), at which point we >lose the info about which part of the estimated phase is due to drift >and which part due to Doppler? So we'd need to keep track of the TX >phase calibrated agains both the fixed delay as well as with a known >relative speed? Honestly, I didn't think too hard about it. :P I was assuming that if one corrects the doppler frequency out with a frequency rotator (not a PLL), that one preserves the TX carrier phase. I'm now questioning that assumption. (I might break out the pencil and paper.) One would need a a very good frequency estimate of the incoming signal's center frequency anyway. >> 4. The patent also mentions one way TOA processing. Using a tx time >> in the message and carefully tracking of received sample time at a >> known point (e.g. the complex cross-correlation peak of the preamble), >> one could perform the TOA estimate, if you have well synchronized >> clocks. Constant known delays through the receiver coaxial cables, >> external filters, etc. would need to be calibrated out. Constant >> delays through the transmitters' coaxial cables, filters, etc. would >> need to be estimated at the receiver and calibrated out. Maybe the >> transmitter can help you out, and send those delays as knowns in the >> measurement message. >I must admit I didn't even go that deep into the patent. From my quick skim, it seemed to be about aggregating a collection of one-way and two-way ranging and navigation techniques, with additional data provided by the two cooperating systems, to figure out relative positions and velocities in an environment where GNSS is not available. My experience with implementing one-way TOA processing with AIS (with GNSS available) taught me that system biases are the majority of the raw measurement. But least those biases are usually constant and can estimated easily and removed with the data coming from the transponders. >>Right now I wish I had the time to go through Information Theory again �C >>in fact, knowing the phase and the frequency shift the TX1->TX2 and the >>reverse channel cause is nothing different from channel state >>information in the very communication theoretical sense, and the >>differential entropy of that source of information must be directly >>related to measurement accuracy/variance one can achieve. I was guessing that two different frequencies were used in the patent for at least 2 reasons: a. it was an FDM communications system between interrogators and transponders b. two separate frequencies might be used to remove Doppler ambiguities Regards, Andy >Best regards, >Marcus _______________________________________________ USRP-users mailing list USRP-users@lists.ettus.com<mailto:USRP-users@lists.ettus.com> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com