Real-time PPP with undifferenced integer ambiguity resolution, experimental results. D. Laurichesse, F. Mercier CNES Orbitography

Transcription

1 Real-time PPP with undifferened integer ambiguity resolution, experimental results D. Laurihesse, F. Merier CNES Orbitography

2 Undifferened ambiguity resolution bakground Identifiation of differential satellites biases Gabor & Nerem (1999), Ge & Al (2007) CNES Method (ION GNSS 2007, ENC-GNSS 2008, ION ITM 2008/2009, Navigation 2009) Satellites biases are absorbed in the loks Appliations to PPP, LEO POD, Real Time integer PPP, CNES/CLS IGS analysis enter (GRG) Collins (NrCan, ION ITM 2008) proposes a similar approah The deoupled lok model, generalization of the CNES method Geng method (University of Nottingham, ION GNSS 2008) Identifiation of UHD (undifferened hardware delays) Similar onepts (Mervart, Roken, GPS Solutions, ION GNSS 2008) Single differene hardware delays and measurements Blewitt (AGU 2009, NBMG & JPL) arrier range onept, biases are estimated after a double differened network solution

3 CNES method overview Wide-Lane (ionosphere-free, geometry-free ombination, 70 m wavelength) ( L L, P P ) = N + hw f w 2 1 1, 2 reeiver - emitter Integer widelane loks phase pseudo-range integer widelane reeiver - emitter pseudo-range loks Narrow-Lane, ionosphere-free ombinations γλ 1 L 1 λ 2 ( L + N ) γ 1 2 w P Q = = D D +λ W + h + h p λ N 1 integer ambiguity Single frequeny ionosphere free system with a wavelength of 10.7 m propagation model 10.7 m reeiver - emitter Integer phase loks Resulting phase loks have integer nature zero-differene ambiguity resolution at reeiver level beomes possible => ambiguity fixed PPP for isolated reeivers

4 Benefits of separate phase and pseudo-range loks SP3 loks: ombine pseudo-range and phase observations Pseudo-range loks smoothed by phase = phase loks aligned on pseudo-range h ( h ) m P, ov P 5 Integer ambiguity model : separate phase and pseudo-range loks h, h p h ov ( h ) ov 1m, h ( h h ) 5 m p Reonstrution of SP3 loks from separate loks is possible (bakward ompatibility) The error on the reonstruted lok is not better than the SP3 one h p = h + ( h h ) p We use the phase lok as the referene - more aurate (pseudo-range measurements are not involved in their omputation) - phase models are better known - integer property

5 Appliation to real-time integer-ppp Test in replay mode PPP-like ompat representation Reeiver Network raw data flow Raw measurements (Nw fixed) widelane fixing (sliding window) Kalman filter (mixed float/integer for N 1 fixing) loks/orbits orretions omputation h w lok GPS orbits h, h p ambiguity fixing strategy based on network onnetivity 1200 parameters for a network of 60 stations and 32 satellites 1000 measurements per epoh loks widelane fixing (sliding window) Kalman filter N 1 fixing and position estimation Raw measurements Raw measurements (Nw fixed) Preise trajetory Network side (orbit and lok estimation) User side (PPP trajetory) 5

6 Impat of fixed ambiguities (Kinemati positioning, replay mode) ION ITM 2009 standard PPP with floating ambiguities ~10 m RMS PPP with integer fixed orbits and loks ~2 m RMS 6

7 Real-time experiment arhiteture (network side) Network Analysis IGS NTRIP CASTERS WEB SERVER Dissemination (2-3 se lateny) FTP BNC X, Y, Z, H, Hw, Hp-H Daily 6 hours WideLane Clok Computation Real-time ontinuous orbit onstrution C1, C2, P1, P2, L1, L2 Hw X, Y, Z, H Real-time (2-3 se lateny) GNSS engine (Kalman filter) Computation: 2 se Overall lateny 6-8 se stations/ satellites onfiguration 7

8 Real-time network of stations NTRIP (use of GFZ aster, Courtesy Georg Weber) 40 stations 1 day of measurements

9 Real-time phase lok solution auray Pure phase loks h Comparison with IGS finals: One ommon offset removed per epoh One onstant offset removed per ontinuous lok segment (user-side phase ambiguity estimation) Correted from radial orbits differenes Errors ~ 1 m (0.03 ns) Muh better than SP3-like loks Cloks have integer property Availability 95 % Global orbits/loks omputation is feasible with urrent Real Time IGS measurements streams

10 Real-time PPP with loal network demonstrator System operational sine Deember 2009 Proesses data from a loal network of 8 Euref stations Adjust position in real-time of a fixed user station GTL1 (Septentrio GeneRx) Réseau de stations utilisées dent ware brst brus graz Latitude ( ) aor ala gtl1 mars x GTL Longitude ( )

11 GTL1 positioning error Comparisons to a referene position 1 m RMS horizontal auray 3 m RMS vertial auray

12 Advantages and drawbaks of the method PPP 2-frequenies RTK 1 or 2 frequenies CNES Integer PPP 2-frequenies Geometry Global Loal (< 50 km) Global Convergene time (TTFF) Horizontal preision Convergene : < 30 m Kik start: 1 mn stati: 15 mn dynami: 30 mn Convergene : ~ 1 m Convergene : ~ 1 m 1-5 mn Kik start: 1 mn stati: 30 mn dynami: 90 mn m ~ 1 m ~ 1 m The method is operationally equivalent to PPP but with the preision of RTK (1 m vs. 10 m) Potential appliation to earthquake monitoring, oil rigs positioning, RTK base station positioning 12

13 Conlusion and future work Worldwide orbits and loks has been omputed in real-time using the urrent RTIGS network and NTRIP tools, results are promising A demonstrator has been working sine Deember 2009 in limited mode Server side: works on a regional network (European-wide) User side: stati reeiver (but proessed in Kinemati mode) Centimeter auray positioning preision Perspetives for 2010 Extension to the global network Proposal to ontribute to the RTIGS projet As a real-time provider of measurements from CNES stations As a real-time analysis enter Improvements of the user-side algorithms Full Kinemati mode proessing Robustness enhanements (for moving reeivers in onstrained environments for example)