The Possibility of Precise Automobile Navigation using GPS/QZS L5 and (Galileo E5) Pseudo ranges

The Possibility of Precise Automobile Navigation using GPS/QZS L5 and (Galileo E5 Pseudo ranges ION ITM ITM 013 Hiroko Tokura, Taro Suzuki, Tomoji Takasu, Nobuaki Kubo (Tokyo University of Marine Scienceand Technology

Outline Background Objective Multipath error estimation methods Static ti and kinematic experiment Test results Conclusion 1

Background GPS L5 up to 4 by 01 L5 is broadcast from 3 satellites (PRN1/4/5 QZS L5 up to 4 by 017 Galileo E5AltBOC :cmlevel accuracyunderopensky under

L5 Signal Tracking by SDR (PRN1 5000 Correlation Value 4000 3000 000 1000 L5 signal Standard L1 C/A GPS PRN1 01/8/16 Open Sky Elevation:45 deg. 10ms Integration FE: Fraunhofer SF: 40.96MHz 0 300 00 100 0 100 00 300 Delay (m 3

Multipath Envelope (hardware simulator check: MP is set 6dB lower than Direct Consumer receiver Pseudo ra ange error rs [m] Geodetic receiver Multipath Delay [m] Consumer receiver is vulnerable to multipath delay even over 100m 4

Multipath Envelope (close up Pseudo range err rors [m] Geodetic receiver Multipath Delay [m] small difference between L1 and L5 Multipath in L1 has been imped 5

Performance in Urban Areas Vehicles in dense urban areas with geodetic receiver DGPS in Tokyo RTK in Tokyo Several applications require high accuracy (RTK far from perfect at present 6

Objective Pseudo range observations from L5 in both GPS and QZS are basically robust against multipath. It can be used for high h reliable and accurate application without Ambiguity Resolution. L5 performance has not been investigated because satellites with L5 are few. The Question: Dose pseudo ranges of new signals really work well in urban areas? 7

Estimation of Multipath Error Code Carrier Carrier difference(cc differnce differnce Effective for static data Cycle slip happens a lot for kinematic data impossible to extract multipath for kinematic data Proposed method Separating Multipath Errors for Vehicle 8

CC difference CC difference L1 = P1 4.0915 x Φ1 + 3.0915 x Φ CC difference L5 =P5 3.51 x Φ5+ 3.51 x Φ1 P : Pseudo range measurement Φ : Carrier phasemeasurement Dual frequency receivers can effectively remove the ionospheric delay Offset average Use both Pseudo range and Carrier phase Effective for only static data 9

Proposed Method Separating Multipath Errors for Vehicle Higher Elevation SV Reference SV Precise Rover Position + Double Difference 4 Different Combinations of Multipath and Noise Code minus Carrier Difference Extraction of Target Multipath and Noise Target Multipath and Noise Precise Rover Position Rover Reference Receiver 10

Separating Multipath Errors for Vehicle P ( mp ( mp ( P c ( dt c( dt c ( dt c( dt noise noise P dt dt dt dt ( P ion ion ion ion ( mp ( mp P tropo tropo noise tropo tropo noise sv sv sv sv 1 mp mp mp mp noise noise noise noise Target = + + Raw Data Measurements 3 1 3 Computed by each CC Difference : Target SV : QZS (Elevation angle > 80 11 11

Proposed Method Separating Multipath Errors for Vehicle Higher Elevation SV Reference SV (QZS is very useful Precise Rover Position + Double Difference 4 Different Combinations of Multipath and Noise Code minus Carrier Difference Extraction of Target Multipath and Noise Target Multipath and Noise Precise Rover Position Rover Reference Receiver 18

Test and Results 1. Static Test 1 (Toyosu,Tokyo Tokyo long distance (approx. 30m from building. Static Test (campas,tokyo short distance (<10m from building 3. Kinematic Test 1 (tukishima, Tokyo 4. Kinematic Test (edagawa, Tokyo Validation for proposed method Estimate multipath error using proposed method Target : GPS PRN1 PRN1 (transmitting both L1 and L5 19

Static Test 1 (Toyosu,Tokyo 1/13/ 01 (GPSTIME 0:30~ Geodetic Receiver 60 min Hz 5 8 satellites in view over 15 degrees elevation Target SV : GPS PRN 1 (L1,L5 Reference SV : QZS 1 (L1,L5 L5 Precise position was computed by post processing Building Rover Receiver 30m W N S Long delay multipath using proposed method to extraction the PRN 1 multipath 0 E

Static Test 1 Multipath Errors (no smoothing [m] 1 0.8 0.6 0.4 0. 0 0. 0.4 0.6 0.8 08 1 347860 348460 349060 349660 35060 350860 351460 GPSTIME [s] [] σ [m] L5 0.37 L1 0.5 1

Static Test 1 Multipath Errors (100s smoothing [m] 1 0.8 0.6 0.4 0. 0 0. We just set receiver parameter as 100s for smoothing 0.4 0.6 0.8 08 1 347860 348460 349060 349660 35060 350860 351460 GPSTIME [s] [] σ [m] L5 0.09 L1 0.1

Static Test (campas,tokyo 1/14/ 01 (GPSTIME 0:30~ N Geodetic Receiver 60 min Hz 5 9 satellites in view over 15 degrees elevation Target SV : GPS PRN 1 (L1,L5 Reference SV : QZS 1 (L1,L5 L5 Precise position was computed by post processing Rover Receiver 9m Building W S Short delay multipath using proposed method to extraction the PRN 1 multipath 3 E

Static Test Multipath Errors (no smoothing [m] 3 1 0 11 σ [m] L5 1.40 L1 1.74 3 433500 434100 434700 435300 435900 436500 437100 437700 GPSTIME [s] 4

Static Test Multipath Errors (100s smoothing [m] 3 We just set receiver parameter as 100s for smoothing 1 0 11 3 433500 434100 434700 435300 435900 436500 437100 437700 GPSTIME [s] σ [m] L5 1.69 L1 1.56 5

[m] 3 1 0 1 Validation adao of Our Proposed Method MP errors derived from cc difference 3 433500 434100 434700 435300 435900 436500 437100 437700 3 1 0 MP errors derived from proposed method 3 433500 434100 434700 435300 435900 436500 437100 437700 GPSTIME [s] Multipath for PRN 1 derived from cc difference match 1 Multipath for PRN 1 derived from Proposed Method 6

[db] 70 60 50 40 30 0 10 Temporal Carrier to Noise oseratio 0 433500 434100 434700 435300 435900 436500 437100 437700 GPSTIME [s] Another type of geodetic receiver shows almost same difference level (4 5dB between L1 and L5 although h the maximum level lwas different. 7

Kinematic Test 1 (tukishima, Tokyo 11/3/ 01 Geodetic Receiver 0 min 5 Hz 5 6 4 8 satellites in view over 15 degrees elevation Target SV : GPS PRN 1 PRN (L1,L5 L5 Reference SV : QZS 1 (L1,L5 Precise position was computed by post processing 3 4 100s smoothing 8 7 Fix rate[%] RTK 74.0 00m (Post Processed RTK Plots 8

Kinematic Test 1 Multipath Errors [m] 100 80 60 40 0 0 0 438000 438300 438600 438900 43900 GPSTIME[s] Enabled percentage[%] L5 49.6 L1 59.3 9

Kinematic Test 1 Multipath Errors [m] 1.5 1 0.5 0 0.5 05 abs(mp < 5m 1 1.5 438000 438300 438600 438900 43900 GPSTIME[s] σ [m] L5 08 0.8 L1 0.77 30

Kinematic Test (edagawa, Tokyo 1/7/01 Geodetic Receiverei er 35min 5 Hz 4 9 satellites in view over 15 degrees elevation Target SV : GPS PRN 1 PRN (L1,L5 L5 Reference SV : QZS 1 (L1,L5 Precise position was computed by post processing 4 3 100s smoothing Fix rate[%] RTK 94.1 00m 31 31

Kinematic Test Multipath Errors [m] 100 80 60 40 Enabled 0 percentage[%] 0 0 430830 431130 431430 431730 43030 43330 43630 43930 L5 64.9 L1 91.5 GPSTIME [s] 3

Kinematic Test Multipath Errors [m] 1.5 1 0.5 0 0.55 1 1.5 430830 431130 431430 431730 43030 43330 43630 43930 abs(mp < 5m σ [m] L5 193 1.93 L1 1.55 GPSTIME [s] 33

Conclusion Pseudo range observables from L5 are basically robust against multipath. We were able to estimate t multipath th errors for moving targets by using the proposed method. The multipath mitigation performance bt between L1 and L5 was not so different at present. Using L5 instead of L1 will be practical in the future without special correlator. 34

Future work Further investigation for L5 signal is required because manufactures are still developing the tracking technique for new L5 signal. Software defined GNSS receiver can be used to evaluate it. 35

Thank you very much for your kind attention!