Tokyo University of Marine Science and Technology Effective Satellite Selection Methods for RTK-GNSS NLOS Exclusion in Dense Urban Environments 15 September 216 Hiroko Tokura, Nobuaki Kubo (TUMSAT) Hitachi Zosen Corporation Geospatial Information Authority of Japan The Geographical Survey Institute carried out this study as a general technology development project of the Ministry of Land, Infrastructure and Transport minister's secretariat technology Security Research Division.
Outline 1. Background and objective 2. Conventional satellite selection methods 3. Testing and results 4. Weakness of SNR and SNR based new method 5. Testing and results 6. Conclusions ION GNSS+ 216 1
Background Multipath effects in dense urban environment DGNSS solutions Google map Testing course Google earth ビルによる回折波 Huge errors Caused by high-rise buildings Multipath effects are problem for GNSS positioning In dense urban environments ION GNSS+ 216 2
Background Two types of multipath effects by NLOS satellites Between the two different height of buildings Signal strength with skyplot Diffraction Low-rise building Diffracted signal Low-rise building Direct signal NLOS signal Reflection High-rise building Reflected signal Results of DGNSS 12hours DGNSS solution High-rise building SNR= 45.. 4.. 35.. 3.. 25 < 25 [db-hz] Extended by multipath signal NLOS signal occurs Multipath errors (Non-line of sight) Mitigate the multipath errors by satellite selection methods 3
Background Increasing number of operational GNSS satellites Increase the number of received satellites by multiple constellation Received satellites by observation data GPS / QZSS / BeiDou / GLONASS Satellite selection to exclude NLOS satellite Improvement of positioning performance SNR= 45.. 4.. 35.. 3.. 25 < 25 [db-hz] LOS 11 NLOS 3 One epoch of actual received signals By the results of experiment 4 Chance to improve positioning performance using satellite selection method ION GNSS+ 216
Background Conventional satellite selection methods The fisheye view image has been used for several researches Suzuki, T., Kitamura, M., Amano, Y., and Hashizume. High-accuracy GPS and GLONASS positioning by multipath mitigation using omnidirectional infrared camera. ICRA 211 Precise 3D building maps are being developed by companies and used for multipath mitigation Hsu, L. T., GU, Y., and Kamijo, S., 3D building model-based pedestrian positioning method using GPS/GLONASS/QZSS and its reliability calculation. GPS Solutions, 1-16.ISO 69 Groves, Paul D., et al. Intelligent urban positioning using multi-constellation GNSS with 3D mapping and nlos signal detection. 212 Images of 3D building These methods are mainly discussed for kinematic data with code based positioning We try to apply these methods for RTK-GNSS Signal strength observation to detect the multipath signal Suzuki, T., Kubo, N., and Yasuda, A., The possibility of the precise positioning and multipath error mitigation in the realtime. In The 24 International Symposium on GNSS/GPS 5
Objective Performance improvement for surveying Target: Multipath mitigation for surveying cm-level positioning (RTK-GNSS) Use of Multi-GNSS Static positioning Evaluation of conventional study of satellite selection method for RTK-GNSS 1. Mask based on fisheye view image 2. Mask based on precise 3D-map 3. Mask based on SNR measurements ION GNSS+ 216 6
Conventional satellite selection methods 1. Fisheye view images based mask Observed signal strength with equidistant projection YASUHARA Co., Ltd. MADOKA18 1 2 3 Procedure for making mask 1. Azimuth adjustment 2. Projection adjustment checkerboard calibrating tools for the initialization 3. Mask Making Binaries the image Projection RTKLIB 2.4.3 b5~ Open source software to make a mask with the fisheye view image SNR= 45.. 4.. 35.. 3.. 25 < 25 [db-hz] Mask: Red line (Expressed by elevation for every 1 deg. Of azimuth) ION GNSS+ 216 7
Conventional satellite selection methods 2. Precise 3D-map based mask Input data Precise 3D map (1cm accuracy) Estimated position by SPP (Several metres) Software Screen shot Output data Sky obstacles mask Sky obstacles comparisons By 3D map By Fisheye view image Input (3D map, position by SPP) Available Input file Expressed Kml same file tendency Shape file By Dr. Suzuki of Waseda Institute for Advanced Study
SNR [db-hz] Conventional satellite selection methods 3. SNR measurement quality check based mask Elevation-SNR estimated line and Threshold line Estimated line Mask line SNR is basically related to the satellite elevation angle 24-hours SNR at base station (Open sky) Elevation [deg] Multipath signal causes a reflection loss 24-hours SNR at rover (Multipath environment) 9
Testing and results Outline of experiments Fisheye view pictures of each testing environment Point A Point B Point C Point D Point E Instantaneous RTK-GNSS (Without any filter, hold technique) Double frequency observations GPS/QZSS/BeiDou Analyse conditions AR: LAMBDA Methods with Ratio test (Fixed threshold for over 3) Elevation mask: Over 15 degrees Short baseline (within 1 Km) Period Receivers Antenna 24hours data at each point A 215-12-9 7:9:3~ 12-1 7:5:3 B 215-12-22 7:53:3~ 12-23 7:53: C 215-12-9 7:9:3~ 12-1 7:9: D 215-12-21 6:54:~ 12-22 6:53:3 E 215-12-21 6:54:~ 12-22 6:53:3 Base / Rover : JAVAD DELTA JAVAD GrAnt-G3T *North side up 1
Testing and results Availability results of each point Availabitliy = Fix solution Total epoch Point A Point B Point C Point D Point E 9 8 9 8 7 7 6 6 5 5 4 4 3 3 2 2 1 1 3.8 [%] 1 9 1 8 9 8 7 7 6 6 5 5 4 28.5 28.5 4 3 28.5 28.5 18. 28.5 28.5 18. 18.3 2 18. 1.1 18. 18. 21.1 3.8 1.1 1.1 3.8 1 1.1 3.8 1.1 13.8 3.8 建物近傍でのRTKのFIX 率 (5つの異なる環境) 建物近傍でのRTKのFIX 率 (5つの異なる環境) 建物近傍でのRTKのFIX 率 (5つの異なる環境) 建物近傍での RTKのRTK FIX の率 FIX (5率つの異なる環境 (5 ) ) 98.5 99.1 96.7 98.5 99.1 建物近傍でのRTKのFIX 率 (5つの異なる環境) 96.7 69.4 83.5 69.483.5 83.5 69.483.5 69.4 69.4 98.5 69.4 98.5 83.5 98.5 98.5 83.5 55.1 55.153.8 46.7 53.846.8 55.1 55.1 46.7 46.8 53.8 46.7 46.8 99.1 99.1 96.7 96.7 99.1 99.1 96.7 96.7 53.8 55.1 62.4 62.4 53.8 62.4 A A A A B B B A A B B BC C C D D E C C C D E D D D E E E E 通常 RTK 通常 Normal 通常 RTK RTK 通常 RTK 通常通常 RTK 魚眼画像を用いたマスク RTK RTK 魚眼画像を用いたマスク Fisheye view 魚眼画像を用いたマスク魚眼画像を用いたマスク 3D 地図を用いたマスク 3D Precise 3D 地図を用いたマスク 3D-map 信号強度観測値劣化判別マスク SNR 信号強度観測値劣化判別マスク Reliability is over the 99% 3D 地図を用いたマスク信号強度観測値劣化判別マスク 55.1 53.8 62.4 62.4 The results of sky obstacles mask by Fisheye and 3Dmap are almost same results Accuracy of 3Dmap and complex shape of the buildings is problems SNR mask is slightly better than fisheye mask 62.4 *There are very few wrong fixing solution 11
Incident angle [deg] Fisheye view mask The important point to make a mask with fisheye view image Lens calibration Checkerboard is used to obtain the Initial calibration value Important points to take a photo Using the camera is difficult to set up to the true north The camera has to be set up at the same place as the antenna with same posture Original photo with observed SNR Equidistant projection model Calibration line by Checkerboards Distance from the image center [pixel] ION GNSS+ 216 12
Fisheye view mask Effects of lens calibration 12hours static data GPS/QZSS/BeiDou/GLONASS Instantaneous RTK-GNSS 1% 8% The results of each calibration model Normal RTK-GNSS Calibrated by Checkerboards checkerboards Calibrated by cos equidistant model No calibration 6% 52.% 45.3% 4% 2% 19.1% 28.7% No calibration Calibrated by equidistant Calibrated by Checkerboards % NLOS exclusion by fisheye view required precise calibration ION GNSS+ 216 13 1
Testing and results Characteristic for the methods 1. Fisheye view mask Density of sky obstacles for both buildings and trees More realistic: same environment as antenna Making mask procedure is manually Initial correction for each lens to adjust projection Not realistic 2. Precise 3D map mask Making masks automatically in advance Trees, distant buildings and complicated shape buildings Depends on accuracy of input position and 3Dmap Limited to the place that exist of precise 3Dmap 3. SNR mask No need for external data Preparation for each estimated line of receiver and satellite systems ION GNSS+ 216 14
Testing and results How is the effect of mitigating for two types of multipath? Diffracted signals by NLOS As a result of previous experiments, diffracted signals can be excluded correctly. Reflected signals by NLOS Because of the building height is almost same, the effect of reflected signal is relatively low. However, there is the situation that received strong reflected signals by NLOS SNR mask is difficult to detect these reflected signals Strong reflected signals are difficult to mitigate We investigated to know the proper performance under this situation Additional experiments were performed ION GNSS+ 216 15
Testing and results at NLOS environments Outline of new experiments Diffraction Specific environment that the receivers force to receive strong reflected signal by NLOS satellites Reflection Antenna Splitter Conditions Instantaneous RTK-GNSS (Without any filter, hold technique) Double frequency observations for GPS/QZSS/BeiDou/GLONASS Testing environment SNR= 45.. 4.. 35.. 3.. 25 < 25 [db-hz] A Powerful reflected signals were contentiously received B Analyse conditions AR: LAMBDA Methods with Ratio test (Fixed threshold for over 3) Elevation mask: Over 15 degrees Short baseline (within 1 Km) Receiver Base/Rover: A, B Diffracted signals are relatively few Satellite selection methods 1. Fisheye view mask 2. SNR mask ION GNSS+ 216 16
建物近傍でのRTKのFIX 率 (5つの異なる環境) 建物近傍でのRTKのFIX 率 (5つの異なる 建物近傍でのRTKのFIX 率 (5つの異なる環境) 建物近傍でのRTK 98.5 99.1 Testing and 98.5 results at NLOS environments 1% 96.7 98.5 99.1 96.7 建物近傍での RTK1 のRTK FIX の率 FIX (5 建物近傍での率つの異なる環境 (5 1 98.5 96.6 RTKのRTK ) FIX の率 ) 建物近傍での FIX (5率つの異なる環境 (5 RTK1 ) のRTK FIX ) の率 FIX (5 建物近傍での率つの異なる環境 (5 RTKのRTK ) FIX の率 ) F 98.5 9 98.5 83.5 98.5 99.1 98.5 83.5 99.1 99.1 99.1 1 96.6 9 98.2 98.6 98.8 96.7 96.7 98.5 9 98.5 83.5 96.7 96.7 98.5 99.1 98.5 83.5 99.1 Availability results of both receivers 8% 1 96.6 98.2 1 96.6 98.2 8 9 8 8 83.5 69.483.5 83.5 69.4 983.5 83.5 9 69.483.5 83.5 69.483.5 8 7 Receiver A 7 6% 55.% Receiver 62.4 8 7 62.4 8 B 69.4 69.4 69.4 69.4 7 6 55.1 69.4 69.4 55.1 69.4 69.4 53.8 53.8 6 7 6 7 62.4 62.4 41.7% 55.1 62.4 62.4 53.8 62. 6 5 55.1 55.153.8 53.8 55.1 55.1 53.8 4% 6 5 53.8 6 5 55.1 55.1 5 53.8 53.8 4 46.7 46.8 5 46.7 4 4 28.5 5 4 4 3 Ave SV 18.2% 28.5 28.5 2% 4 Ave SV 28.5 28.5 28.5 3 3 4 18. 3 28.5 28.5 28.5 28.5 2 All 11.4 3 18. All 12.4 18. 28.5 28.5 18. 18. 2 1.1 3 2 18. 18. 18. 1.1 18. 1.1 1 1.1 3.8 GJ 1.14.3 2 2 % 1 GJ 4.7 14 18. 18. 1 1.1 3.8 1.1 2 1.1 3.8 3.8 8 1.1 C 3.8 1 1 1.1 3.8 3.8 3.8 1.1 1 C 3.8 13.8 3.8 R A3.3 B C D R 3.8 A A B B C C D D E AE E A A B B B SNR= 45.. 4.. 35.. 3.. 25 < 25 [db-hz] C A1% A B B B 通常 CRTK 通常 Normal 通常 CRTK RTK RTK C D D D 魚眼画像を用いたマスク Fisheye 1% AE view E魚眼画像を用いたマスク E A A B B B 通常 CRTK 通常 Normal 通常 CRTK RTK RTK C D D D RTK 通常 Normal 通常 RTK RTK RTK 3D 地図を用いたマスク 3D 魚眼画像を用いたマスク Precise Fisheye 3D 地図を用いたマスク 3D-map view 魚眼画像を用いたマスク通常 RTK 通常 Normal 信号強度観測値劣化判別マスク通常 C/NRTK RTK RTK 信号強度観測値劣化判別マスク 3D 地図を用いたマスク 3D 魚眼画像を用いたマスク Precise Fisheye 3D 地図を用いたマスク 3D-map view 魚眼画像を用い 3D Precise 3D 地図を用いたマスク 8% 3D-map 信号強度観測値劣化判別マスク SNR 信号強度観測値劣化判別マスク 3D 8% 3D 地図を用いたマスク 3D Precise 地図を用いたマスク 3D-map 信号強度観測値劣化判別マ SNR 6% 4% 2% 19.1% 52.% 4.4% 6% 4% 2% 8.8% 33.9% 23.3% % % As expected, Fisheye 1 view mask is more efficient to exclude multipath signal 1 We improved SNR mask based on the fisheye view mask.. ION GNSS+ 216 17
Testing and results at NLOS environments Remaining SNR observations of reflected signal SNR= 45.. 4.. 35.. 3.. 25 < 25 [db-hz] SNR mask Lots of strong reflected signals were remained Strong reflection signal Observed SNR Applying SNR mask Diffracted signals were removed The remaining SNR was analyzed based on fisheye view mask NLOS signal remained Time series of SNR NLOS LOS (Analyse by fisheye mask) Conventional SNR mask cut off lower SNR below the line Improved satellite selection method focused on variation strong variation are appeared by reflected signals ION GNSS+ 216 18
Testing and results at NLOS environments Proposed new SNR based satellite selection methods v t i = SNR t i ele SNR ele (1) V(t i )= N 1 (v(t N i )) 2 i=1 (2) Disturbance appeared N is the averaging window size. 1. Take the difference between Estimated SNR line and observed SNR (1) 2. Calculate the backward moving average over the N epoch (2) Threshold line Huge SNR degradation is able to be distinguished Effectively for continuously received reflected signal ION GNSS+ 216 19
Testing and results at NLOS environments 8% New results of proposed method 6% 55.% 5.8% 41.7% Receiver A 4% Receiver B 1% 8% 6% 4% 2% 19.1% 52.% 5.7% 4.4% Normal RTK Fisheye view SNR New SNR 2% % 1% 8% 6% 4% 2% 18.2% 8.8% SNR= 45.. 4.. 35.. 3.. 25 < 25 [db-hz] SNR mask New SNR mask SNR mask New SNR mask 1 Normal RTK Fisheye view SNR New SNR 33.9% 23.3% 28.4% % % ION GNSS+ 216 2
Conclusion 3 methods were evaluated at the static positioning Sky obstacles mask by precise 3D-map showed almost the same performance as a fisheye view mask The SNR based mask is the powerful and effective method to remove the quality deterioration signal Availably results of applying conventional methods are improved more than 2 times Additional experiments for the strong reflected signal As expected, fisheye view exclusion improved powerfully than SNR New SNR mask was proposed to refer the fisheye view mask The proposed SNR mask is able to be excluded strong reflected signal ION GNSS+ 216 21
Thank you for your attention! ION GNSS+ 216 22
Background Two types of multipath effects by NLOS satellites Between the two different height of buildings Signal strength Diffraction Low-rize building Diffract signal Low-rize building Direct signal NLOS signal Reflection High-rize building Reflect signal SNR= 45.. 4.. 35.. 3.. 25 < 25 [db-hz] Multipath errors in Pseudorange 12hours DGNSS solution High-rize building Mostly affected by reflect signal Multipath occurs by NLOS signal (Non-line of sight) Reflect signal by NLOS satellite is difficult to mitigate Satellite selection to exclude NLOS is effective 23
Testing and results Number of satellite comparison L1, B1 (GPS/QZSS/BeiDou) SNR on SKYPLOT Observed SNR= 45.. 4.. 35.. 3.. 25 < 25 [db- Hz] BeiDou GPS+QZSS Fisheye view mask BeiDou GPS+QZSS 14 12 1 8 6 4 2 14 12 1 8 6 4 2 NLOS satellites 14 14 14 12 12 12 1 1 1 8 8 8 6 6 6 4 4 4 2 2 2 A A A BA BA B CB CB C DC DC D ED ED E E E 24
Point C Point A Testing and results Fisheye mask and SNR mask comparison (L1, B1) 地点地点 B A A Point 地点地点 C B C L1, B1 (GPS/QZSS/BeiDou) SNR on SKYPLOT B 地点地点 D C C 地点 D ED 地点 EE Point A Observed Fisheye view mask SNR mask 建物近傍での建物近傍での RTK 建物近傍でののFIX RTK 率 (5 のつの異なる環境 RTK FIX RTK 率の (5 FIX FIX つの異なる環境率 (5 (5 つの異なる環境 ) ) ) [%] 建物近傍での RTKのFIX 率 (5 RTK つの異なる環境 ) 建物近傍でののFIX 率 (5つの異なる環境 RTK RTK ) のFIX FIX 率つの異なる環境 ) )) 98.5 98.5 98.5 99.1 99.1 99.1 1 [%] 98.5 96.799.1 96.7 98.5 98.5 99.1 99.1 96.6 98.2 96.7 98.6 98.8 96.7 98.5 1 RTK FIX FIX (5 96.6 (5 98.2 98.6 98.8 建物近傍での 1 建物近傍での RTKのRTK FIX の率 FIX (5率つの異なる環境 (5 98.5 ) ) ) 96.7 ) 99.1 96.6 98.2 98.6 96.7 98.8 物近傍での 1 建物近傍での RTKのRTK FIX RTK の率 FIX (5 99.1 96.6 のRTK 率つの異なる環境 98.2 FIX の (5 率 FIX 98.6 (5率つの異なる環境 (5 ) 98.8 96.7 ) ) ) 96.7 9 RTK 98.5 のRTK 98.5 FIX 83.5 の率 FIX 98.5 (5率つの異なる環境 98.5 9 (5 RTK 98.5 98.5 FIX 83.5 FIX (5) 99.1 (5) 99.1 ) 99.1 ) 96.7 9 1 建物近傍での 83.5 99.1 99.1 99.1 RTK 98.5 のRTK 98.5 FIX 83.5 の率 FIX (5率つの異なる環境 KIX の率 FIX (5率つの異なる環境 (5 ) ) 96.7 99.1 96.7 ) 99.1 ) 96.7 9 1 96.6 98.2 96.7 98.6 96.7 98.8 96.7 96.7 98.5 98.5 83.5 98.5 98.5 83.5 99.1 99.1 99.1 99.1 8 99.1 98.5 99.1 98.5 83.5 96.7 99.1 96.7 99.1 96.7 96.7 1 89 8 96.7 96.7 96.7 96.7 83.5 69.483.5 83.5 69.483.5 9 89 83.5 69.483.5 83.5 69.483.5 9783.5 83.5 69.483.5 8 69.483.5 83.5 69.483.5 62.4 62.4 62.4 8 7 8 62.4 69.4 7 69.4 69.4 69.4 69.4 69.4 86 69.4 69.4 55.1 62.4 62.4 7 55.1 55.1 53.8 53.8 62.4 69.4 69.4 69.4 69.4 7 6 7 6 69.4 69.4 55.1 Observed 53.8 62.4 55.1 55.1 62.4 62.4 53.8 62.4 62.4 62.4 7 6 53.8 53.8 6 5 55.1 62.4 62.4 55.1 55.1 62.4 55.1 62.4 53.862.4 62.4 53.8 53.8 6 5 55.1 55.1 55.153.8 55.1 53.8 53.8 53.8 62.4 62.4 6 5 46.7 53.8 55.1 46.8 53.8 53.8 55.1 55.153.8 53.8 55.1 55.1 6 5 53.8 53.8 5 46.7 46.8 46.762.4 46.846.762.4 46.7 46.855.1 55.1 4 53.8 53.8 5 4 55.1 55.1 5 46.7 46.8 4 4 28.5 28.5 5 28.5 53.8 53.8 28.5 28.5 46.73 4 46.8 28.5 4 346.7 46.8 28.5 3 28.5 28.5 28.5 4 28.5 28.5 3 18. 18. 28.5 28.5 28.5 28.5 3 2 3 18. 18. 28.5 28.5.1 18. 2 3 2 8. 18. 18. 2 18. 1.1 18. 1.1 2 1.1 18. 18. 2 1 2 18. 1.1 3.8 18. 1.1 1.1 1 23.811.1 1.1 3.8 1.1 1 3.8 1 3.8 1.1 3.8 1.1 3.8 1.1 3.8 3.8 1.1 1 1 3.8 3.8 3.8 1 3.8 Fisheye view mask B B B B A B C C C C C D ED E A AB A AB D D D E E B B B ABA BC B C CB E E E 通常 RTK C C BC B DC DC C CD D C D D C ED E D D E E D E D E E E E E RTK RTK A TK 通常 RTK 通常通常通常 A RTK RTK B 魚眼画像を用いたマスク B 魚眼画像を用いたマスク C E C D D E E 通常 RTK 通常 Normal 通常通常 RTK RTK 通常 RTK 通常通常 RTK RTK 魚眼画像を用いたマスク Fisheye view 魚眼画像を用いたマスク魚眼画像を用いたマスク魚眼画像を用いたマスク魚眼画像を用いたマスク通常 RTK 通常 RTK 信号強度観測値劣化判別マスク魚眼画像を用いたマスク 3D 図を用いたマスク地図を用いたマスク C3D Precise 3D C3D 地図を用いたマスク 3D-map C 3D 3D 地図を用いたマスク 3D 3D 3D 3D D D D 地図を用いたマスク信号強度観測値劣化判別マスク信号強度観測値劣化判別マスク C/NE 信号強度観測値劣化判別マスク E E 信号強度観測値劣化判別マスク信号強度観測値劣化判別マスク魚眼画像を用いたマスク Fisheye view 魚眼画像を用いたマスク under this situation 信号強度観測値劣化判別マスク SNR 信号強度観測値劣化判別マスク SNR mask SNR= 45.. 4.. 35.. 3.. 25 < 25 [db-hz] Clearly degraded SNR was removed by SNR mask ION GNSS+ 216 25