Framework and Performance Evaluation of a Ray Tracing-Software Defined Radio Method for GNSS Positioning in an Urban Canyon Environment

Transcription

1 17 Nov. 2017, ITSNT2017 Framework and Performance Evaluation of a Ray Tracing-Software Defined Radio Method for GNSS Positioning in an Urban Canyon Environment Rei Furukawa, Tokyo University of Marine Science and Technology / Kozo Keikaku Engineering, Inc. Yoto Emori, Yoshimi Fujii, Yukiko Kishiki, Kozo Keikaku Engineering, Inc. Takuji Ebinuma, Chubu University Nobuaki Kubo, Tokyo University of Marine Science and Technology rei-furukawa@kke.co.jp 1

2 Table of Contents 1. Background 1. GNSS multipath error in urban area 2. Raytracing approach 3. Software Defined Radio approach 2. Framework of Ray Tracing Software Defined Radio method 3. Preliminary Evaluation 1. Static Scenario 2. Moving Scenario 4. Summary 2

3 Background of this work Evolution of GNSS High Accuracy Positioning Popular in construction, agriculture, survey Not popular in city area s positioning Urban canyon environment affects High Accurate Positioning RTK FIX! RTK FIX! RTK FIX! RTK NOT FIX! Opensky Urban canyon We are developing simulator for GNSS users who want to know availability of High Accurate Positioning in urban areas GNSS receiver researchers and developers doing research algorithm for multipath environment 3

4 Background of this work Example of GNSS positioning error in urban area Positioning results Actual route Vehicle positioning (Trimble NetR9+RTKLIB single). Maximum horizontal error 100m 100m Generally it is caused by multipath errors HIBIYA, TOKYO, JAPAN Deep Urban area*1 Why multipath affects GNSS positioning errors? *1:Aerial photo: Geographical Survey Institute Map KML 4

5 Background of this work GNSS multipath error in urban area Reflect Direct Reflect Blocked LoS Satellite Tracking direct signal and reflected signal causes tracking error (depends on algorithm and correlator) NLoS Satellite Tracking NLoS reflected signal causes large tracking error (depends on reflected path length) To analyze what happens in an urban area, we developed GNSS Raytracing multipath simulator. 5

6 Raytracing approach Raytracing multipath simulation *1 with 3D buildings Propagation Path Type 1 Diffraction&1 Reflection 1 Diffraction 1 Reflection 2 Reflections Satellites position by ephemeris All types of propagation paths done by raytracing. Raytracing used to estimate received level and delay from satellites to GNSS antenna 1 Level 1 Reflection Diffraction MultipathProfile Delay Receiver processing model HIBIYA,TOKYO,JAPAN SNR Multipath error of pseudorange Positioning calculation Aerial photo: Geographical Survey Institute Map KML *1 Our raytracing simulation accuracy is validated in "Analysis of Path Gain Inside Tunnels Based on FDTD and Ray Tracing Methods.," 2013 International Symposium on Electromagnetic Theory, G. S. Ching, K. Tsuda, Y. Kishiki, 6

7 Raytracing approach Example scenario of Raytracing Vehicle positioning and recorded in Hibiya. simulated signal quality check of urban environment with measurement data Setting Satellite System Value GPS+QZS L1 3Laps 走 Duration Satellite position Receiver Position Multipath Positioning Method GNSS Receiver 30 minutes Ephemeris Vehicle position (POSLV) Raytracing with 3D buildings(hibiya) Single(RTKLIB v2.4.2) Trimble Net R9 Aerial photo: Geographical Survey Institute Map KML 1.5km Measurement course (with POSLV & NetR9) 7

8 Raytracing approach GNSS SNR simulation in urban area SNR trend of raytracing and actual measurement for G23 SNR[dB] Lap1 Lap2 Lap3 Simulation Raytracing Measurement Tree Obstruction Tree Obstruction Tree Obstruction :28:08(GPST) 23:57:09 time[s] Time[s] Good simulation performance with 3D simple building model*. Signal availability has almost the same trend between simulation and actual measurement. Actual measurement s trend is more dynamic than simulation result. *Accuracy of building height is ±1.5m. Material is concrete. Model has no tree and signboard etc. 8

9 Raytracing approach GNSS SNR simulation in urban area SNR trend of raytracing and actual measurement for G23 SNR[dB] Lap1 Lap2 Lap3 Simulation Raytracing Measurement These fading in measurement on each lap occurred at the same location. (difference with simulation due to objects not included in model.) Time[s] Differences between raytracing and measurement occurred due to tree obstruction. Simulating tree effects is our future work in raytracing. Further improvement is expected. *Accuracy of building height is ±1.5m. Material is concrete. Model has no tree and signboard etc. 9

10 Raytracing approach GNSS SNR simulation in urban area Simulation accuracy depends on Raytracing engine 3D model s accuracy many 3D model vendor in Japan SNR simulation in urban area in Japan is possible How about GNSS positioning accuracy simulation? 10

11 Raytracing approach GNSS positioning simulation in urban area RTKLIB single positioning with raytracing and actual RINEX obs similar trends by raytracing are obtained Raytracing Positioning Result Large errors Small errors Actual Positioning Result Large errors Small errors Positioning result depends on receiver processing algorithm. How to estimate consumer GNSS receiver performance in urban area? Aerial photo: Geographical Survey Institute Map KML 11

12 SDR approach Algorithm of consumer GNSS receiver is unknown (black box) Difficult to get positioning performance of consumer GNSS receiver through numerical simulation Emulate multipath GNSS signal is a good way Raytracing 1 Level 1 Reflection Diffraction 2 3 Level 1 MultipathProfile 2 Delay 3 2 α MultipathProfileβ Delay delay1 3 delay2 GNSS signal Generation LoS TxSignal1 + MP TxSignal2 - MP TxSignal3 MultiPath TxSignal Multipath included GNSS Signal consumer α GNSS receiver β Positioning result Traditional signal generator is an excellent solution, but it is expensive. We focused on Software Defined Radio. 12

13 SDR approach Software Defined Radio Wireless system built with software digital signal processing and general purpose RF-frontend (low cost) gps-sdr-sim(original) RF-Frontend ($500) Satellite Position calculation GPS signal generation modulation USB3.0 I-Q sample Amplifier, D/A, A/D, Filter RF cable GPS Signal GNSS Receiver Hardware control SDR signal generator needs only one RF channel for all GPS signal. It is the same for many multipath channels. 13

14 Framework of RayTracing Software Defined Radio(RT-SDR) Method Overview RayTracing 1 Level 1 Reflection Diffraction MultipathProfile Delay Software Defined Radio gps-sdr-sim(customized)* -support QZSS signal generation -Troposphere, Ionosphere emulation -Multipath signal Generation -GPU: for many multipath calculation Multipath contained GNSS Signal RF cable *we named customized gps-sdr-sim SDR-SAT GNSS Receiver Now, Our prototype, GPS L1 and QZS L1 are available. 14

15 Preliminary Evaluation We used two types of scenario for RT-SDR prototype Static Scenario Emulated signal quality check of SDR signal generator with simple scenario Signal quality check with RTK fixing rate Moving Scenario Emulated signal quality check of urban canyon environment Signal quality check with measurement data 走行コ Measurement course (with POSLV & NetR9) *1 15

16 Static Scenario Scenario setup Scenario Settings Setting Value Satellite System GPS+QZS L1 Duration 1 Hour Satellite position Ephemeris Receiver Static Position Multipath None(open sky) Positioning Method RTK continuous (RTKLIB v2.4.2) GNSS Receiver u-blox NEO-M8T SDR Settings Setting Value Frequency [GHz] RF Frontend Blade RF x40 Sampling rate 26 [MHz] OS Windows 10 64bit CPU Core i GPU NVIDIA GeForce GTX

17 Static Scenario N Evaluation Result Float solution Fix solution S Skyplot of All Satellites Fixing rate=60% Some miss Fix solutions are observed. Time[s] We need signal quality improvement in GNSS phase simulation. 17

18 Moving Scenario Scenario setup Setting Satellite System Duration Satellite position Receiver Position Multipath Positioning Method GNSS Receiver Scenario Settings Value GPS+QZS L1 30 minutes Ephemeris Vehicle position (POSLV) Raytracing with 3D buildings(hibiya) Single/DGNSS/RTK continuous (RTKLIB v2.4.2) u-blox NEO-M8T SDR Settings Setting Frequency RF Frontend Sampling rate OS Value [GHz] Blade RF x40 26 [MHz] Windows 10 64bit CPU Core i GPU NVIDIA GeForce GTX1080 used Net R9 for actual measurement We will use future evaluation. 18

19 Moving Scenario satellite evaluation result for G23(NLoS) Skyplot of G23 SNR[dB] 55 RT-SDR Actual :28:08(GPST) time[s] 23:57:09 Signal attenuation is emulated, and some multipath fadings are emulated. 19

20 Moving Scenario satellite evaluation result for G23(NLoS) SNR[dB] Lap1 Lap2 Lap3 RT-SDR Actual These fadings are not included in current simulation model. (Tree obstruction.) Skyplot of G :28:08(GPST) time[s] 23:57:09 Further improvement is expected with raytracing model s updates. 20

21 Moving Scenario satellite evaluation result for G09 (LoS) Skyplot of G09 SNR[dB] RT-SDR Actual :28:08(GPST) time[s] 23:57:09 Differences of SNR between RT-SDR and actual measurements are observed. 21

22 Moving Scenario Single Positioning Result RT-SDR Single Actual Single Some differences on single positioning result between RT-SDR and actual measurements are observed. We will evaluate again with updated prototype. 22

23 Moving Scenario DGPS Positioning Result RT-SDR DGPS Actual DGPS Many differences on DGPS positioning result between RT-SDR and actual measurement observed. We will evaluate again with updated prototype. 23

24 Moving Scenario RTK Positioning Result No RTK FIX solution from both method due to insufficient number of satellites Some miss FIX solution observed For evaluation of urban canyon situation, we need more GNSS system emulation function. 24

25 Summary and future work Summary We developed RT-SDR framework and prototypes RT-SDR method have many possibilities complex multipath environment emulation build evaluation system with low-cost hardware Future work RT-SDR method have many challenges Quality improvement GNSS signal emulation quality of SDR» Time synchronization to GPST Multipath signal emulation quality of RT-SDR Functional improvement Multi GNSS : GLONASS/BEIDOU/GALILEO Multi frequency : L2, L5, L6 25