Reduction of Pseudorange Multipath Error in Static Positioning Tokyo University of Mercantile Marine Nobuaki Kubo Akio Yasuda
Brief Many researchers have tried to reduce the multipath effect from both hardware and software. Due to their efforts, long-delay code multipath is significantly reduced and carrier multipath also can be reduced recently. Short-delay code multiapth is still a problem. My purpose is to reduce short-delay code multipath in a static positioning.
DGPS under multiapth condition Height (m) 84 83 82 81 80 79 78 77 290000 290500 291000 291500 292000 GPSTime (s) Non Multipath Estimation, Non Smoothing *Novatel RT-2 *Mask 5 degrees *Csmooth 20s *Model502 Ant. (choke-ring) Reference station was installed under clear condition. Rover station was installed under multipath condition. Next picture shows the environment around rover station.
C/N0 Code Multipath (m) 48 46 44 SV8 42 40 38 36 SV22 34 32 30 SV27 290100 290600 291100 291600 292100 GPSTime (s) 10 5 0 290100 290600 291100 291600 292100-5 -10 SV22 : 0m offset SV8 : +7m offset SV27 : -7m offset GPSTime (s) This picture shows C/N 0 in rover station. Main reason SV8: Reflection by Wall SV22: Reflection by BWO SV27: Diffraction This picture shows Code Multipath in rover station. BWO: Box for Weather Observation
Purpose Observing effects of code multipath due to near obstruction. Estimating code multipath due to near obstruction as correctly as we can. Improving accuracy in case of shortdelay code multipath. (Short period and long period test)
Multipath effect on pseudorange (1) In principal, code multipath can be estimated if we can parameterize signal amplitude, time delay, and phase of multipath. When GPS rover antenna moves, it is very difficult to estimate them. Because multipath parameters vary according to the environments. But in case of stationary antenna, we can almost estimate multipath parameters.
Multipath effect on pseudorange (2) Multipath error 1.0 0.4-1.0 0 Direct signal Reflected signal Time delay(chip) Correlation peak with constructive interference from a reflected signal 0.7 1.0 This picture shows correlation peak contaminated by a reflected signal. Most GPS receivers use a DLL (early-late correlator) to track the signal. 30 Multipath error (m) 20 10-10 -20-30 0.1chip 1.0chip Relative Multipath Amplitude = 0.25 100 200 300 400 This picture shows bounds on C/A code error due to multipath. Differential path length (m)
Target Specular code multipath L1-L2 GPS receiver (carrier phase output) Static DGPS Code-minus-carrier (m) 40 35 30 25 20 15 10 5 0 290100 290600 291100 291600 GPSTime (s) GPS Antenna was installed near the wall (3.5m) 30minutes data SV with many cycle-slips are removed
Steps for estimating code multipath Detect cycle-slips. Determine cycle-slip free interval each SV. (Max 30 minutes) Calculation of zero-mean code multipath from code-carrier technique. (A) Estimation of multipath parameters from Satellite-GPS Antenna geometry. Determine mean code multipath. (B) New code multipath is estimated from both A and B.
Method to estimate code multipath Variation is almost correct, but mean can t be estimated. Estimation of multipath parameters Code-minus-carrier technique + Mean code multipath New code multipath
Estimation example Code-minus-carrier (m) 2 1.5 1 0.5 0-0.5 465750 466250 466750-1 -1.5-2 GPSTime (s) Code multipath (m) 2 1.5 1 0.5 0-0.5 7.7 7.8 7.9 8 8.1 8.2-1 -1.5-2 delay (m) 0-mean code multipath variation from code-carrier technique Code multipath from multipath parameters. Amplitude=0.15 New code multipath = Variation of code multipath + Mean code multipath
DGPS TEST Rover station is 3.5m from the wall. Reference station is installed under clear condition. Applying new code multipath to pseudoranges in rover station. Short and long period. RT-2 GPS Receiver and Model 502 Antenna Wall BWO Funnel (choke-ring) * Rover station 10m Reference station * Rooftop Antenna Configuration
Steps for DGPS Produce correction data from measurements of reference station Produce new code multipath in rover station Produce new pseudorange in rover station. (raw pseudorange + new code multipath + correction) Calculate the rover position by least square method
Short period test (1) Height error (m) Height error (m) 84 82 80 78 76 290100 290600 291100 291600 292100 GPSTime (s) 84 82 80 78 76 290100 290600 291100 291600 292100 GPSTime (s) No multipath estimation No carrier smoothing 6m (max variation) 1.19m (height 1σ) Multipath estimation No carrier smoothing 2m (max variation) 0.40m (height 1σ) Precise height : 81.00m
Short period test (2) Height error (m) 84 82 80 78 76 290100 290600 291100 291600 292100 GPSTime (s) No multipath estimation Carrier smoothing (100s) 3m (max variation) 0.74m (height 1σ) Height error (m) 84 82 80 78 76 290100 290600 291100 291600 292100 GPSTime (s) Multipath estimation Carrier smoothing (100s) 0.5m (max variation) 0.10m (height 1σ) Precise height : 81.00m
Long period test MPE: Multipath Estimation 1) Antenna-Wall (3m, 15hours) Pattern Height Horizontal Average No MPE 1.12m 0.76m 81.22m MPE(zero-mean) 0.57m 0.25m 81.21m MPE(+mean) 0.40m 0.22m 81.18m 2) Antenna-Wall (7m, 15hours) No MPE 0.70m 0.55m 81.09m MPE(zero-mean) 0.27m 0.23m 81.19m MPE(+mean) 0.19m 0.17m 81.05m DGPS accuracy (1σ) on MP estimation methods
Conclusion Code minus carrier technique can approximately estimate code multipath. Mean code multipath can be calculated by estimating multipath parameters and accuracy is a little better than only code multipath. We will apply this new technique to correction generating in reference station. The future object is to build technique to estimate code multiapth in real-time and to enhance robustness of this technique.