A Combined Ray Tracing Simulation Environment for Hybrid 5G and GNSS Positioning Ivana Lukčina , Phuong Bich Duonga , Katrin Dietmayera , Sheikh Usman Alib , Sebastian Krama , Jochen Seitza and Wolfgang Felbera a Fraunhofer Institute for Integrated Circuits (IIS), Nordostpark 84, Nuremberg, Germany b Technische Universität München (TUM), Arcisstraße 21, Munich, Germany Abstract GNSS based radio frequency (RF) positioning has to cope with challenging propagation conditions, like non-line of sight (NLoS), multipath, and sparse signal availability. The introduction of the fifth- generation of mobile telecommunications technology (5G) with an improved Positioning Reference Signal (PRS) structure will be a key enabler for more reliable positioning solutions with increased availability and advanced signaling. Nevertheless, 5G-assisted positioning faces similar challenges. Therefore, to analyze the possibilities of 5G-assisted positioning, a suitable simulation environment is required. In this paper, a simulation environment based on a Ray Tracing (RT) channel model that emulates Global Navigation Satellite System (GNSS) signals is introduced, validated and extended to simulate 5G PRSs, and Sounding Reference Signals (SRSs). Additionally, the environment is ap- plied for hybrid positioning by sensor data fusion with real-world recorded Global Positioning System (GPS) L1CA and Galileo E1BC GNSS signals under several severe conditions like strong building block- age and outdoor-indoor transition. It is shown that the simulation environment with various three- dimensional (3D)-modeled objects represents 5G signals sufficiently well when the line of sight (LoS) is visible. Additionally, the simulated 5G signals improve the GNSS positioning accuracies when com- bined in a hybrid positioning approach, especially under complex channel conditions, like in typical industrial environments. Keywords GNSS, GPS, Galileo, 5G, hybrid positioning, Ray Tracing, simulation environment 1. Introduction The investigation and comparison of different simulation environments are relevant to evalu- ate different position fusion algorithms and their performance. Simulation opens up the pos- sibility to meaningfully analyze new signal structures, specific frequency bands, chosen sig- nal bandwidths, and different environmental conditions. This has direct influence on achiev- able positioning performance. Various use cases have already been analyzed using determin- istic, geometric-based stochastic channel models (GSCMs) and non-geometrical stochastic ICL-GNSS 2021 WiP Proceedings, June 01–03, 2021, Tampere, Finland " ivana.lukcin@iis.fraunhofer.de (I. Lukčin); phuong.bich.duong@iis.fraunhofer.de (P.B. Duong); katrin.dietmayer@iis.fraunhofer.de (K. Dietmayer); sheikh.ali@tum.de (S.U. Ali); sebastian.kram@iis.fraunhofer.de (S. Kram); jochen.seitz@iis.fraunhofer.de (J. Seitz); wolfgang.felber@iis.fraunhofer.de (W. Felber)  © 2021 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0). CEUR Workshop Proceedings (CEUR-WS.org) CEUR http://ceur-ws.org Workshop ISSN 1613-0073 Proceedings channel models provided in [1]. The channel model parameters for GSCMs are based on ex- haustive and complex measurement campaigns. For non-geometrical stochastic models, the channel is modeled stochastically with predefined parameters for the environment [2]. In this paper we focus on presentation and evaluation of a RT based simulation environment, to highlight the necessity of realistic position evaluation, with two main signal types: GNSS and sub-6GHz 5G (e.g. 5G FR1). The verification of 5G signals is based on emulated, and GNSS signals on real-world data comparing the signal-to-noise ratio (SNR) and carrier-to- noise density ratio (C /N0 ) values, respectively. Finally, a simple position fusion is done using simulated 5G and recorded real-world GNSS data. 2. Problem description Based on the overview of 5G positioning scenarios and use cases in [3], high mobility, urban area, location awareness for Internet of Things (IoT) applications in urban areas, and indoor- outdoor transition are the ones that should profit from hybridization of 5G and GNSS signals. In high mobility cases, GNSS signals will handle mobility and coverage, while 5G New Ra- dio (NR) can improve accuracy. The opposite occurs for indoor-outdoor transition where 5G NR access points ensure signal availability indoors. For all those use cases, the challenges like GNSS satellite visibility, multipath on both 5G and GNSS signals, obscured or absent LoS components demand further analysis. To assess sensor data fusion challenges properly, it is desirable that different RF signal types undergo the same environmental conditions. Fur- thermore, it is necessary to fully simulate GNSS signals in a specific environment to access the data at different processing stages. Here, the use of real GNSS data should be considered if an accurate positioning reference system is available. Representative simulation should be made available as a preceding step to save time and effort otherwise invested in the field- trials preparations, costs and executions. 3. Method Fig. 1 shows the simulation and measurement setup for real/emulated GNSS and simulated 5G data. The Leica total station is used as a reference system [4]. The GNSS measurement setup was composed of a Septentrio receiver, and a dual circularly polarized GNSS antenna (GNSSA DCP), developed by Fraunhofer IIS and distributed by TeleOrbit GmbH [5]. The Septentrio receiver logging was used to compare the emulated and measured data of some selected pseudorandom noises (PRNs) (Fig. 3a) and as a source of GNSS measurements for positioning. For emulation purposes, Sim3D is used, a GNSS signal emulator that couples a Spirent Signal Generator together with the SE-NAV RT channel model to generate realistic, environment-dependent GNSS signals [6]. The 5G emulation setup is described in [7], where Universal Software Radio Peripheral (USRP) based transmission points (TRPs) are used for 5G signal transmission. The 5G TRP deployment was reconstructed in SE-NAV together with cor- responding trajectories as depicted in Fig. 2c. The perfect synchronization between the TRPs is assumed. To simulate 5G signals, their positions, the signal carrier frequency (3.75 GHz), signal bandwidth (100 M H z), transmitting power (0.032 W), and antenna pattern are the MEASUREMENT SIMULATION Sim3D SE-NAV-EXT SE-NAV SimGen Ranging-Pkg - 4G signal Reference System - indoor-outdoor scenario - 5G signal - Leica - industrial scenario channel data, multipath link - GNSS signal - urban scenario budget data, receiver data, - UWB signal transmitter data Reference Data TOA measurements, RF signal BS positions RF signal GNSS GNSS RX Position Fusion Logging File Figure 1: Measurement and simulation setup (a) Fraunhofer IIS building (b) Modeled IIS building (c) 5G TRPs deployment Figure 2: Real and modeled IIS building with L.I.N.K. test and application center in Nuremberg most important settings that determine the ray properties and received signal power. The transmitter antenna model is defined as 120-degree opening angle antenna. The basis for the simulation is the 3D-model of the Fraunhofer IIS building. Figs. 2a and 2b show real and rebuilt building, respectively. The goal was to have a proper simulation environment of the backyard since the tests were performed in front of the L.I.N.K. test and application center, between two buildings and indoors (Fig. 2c). The simulated building shows a roof made of glass only for the visibility of the indoor area. The used RT channel model [6] provides the option to specify the material permittivity, conductivity, and thickness which define the properties of reflected, dispersed, and transmitted rays (Fig. 2b). For signal propagation: a maximum of 2 reflections and 2 transmissions with enabled diffractions were simulated. Rays with more then 2 reflections or transmissions are neglected. Fig. 2c visualizes those LoS, re- flected, diffracted and transmitted ray paths in white, red, blue and green, respectively. Since SE-NAV is a pure channel model, added software (SW) extensions enable proper signaling for the fusion of different measurements. The developed ranging package includes various sig- nal generators for signals like 5G and GNSS. Scene-dependent ranging signals are generated based on the extracted RT channel properties. 4. Evaluation First we verify the 3D-model based on GPS measurements. The building model shown in Fig. 2c has the same orientation as the skyplot in Fig. 3a. Figs. 3b and 3c show the evaluations GPS L1 GPS L1 60 60 0° 0° GPS 330° 30° 50 50 G15 30° 300° G20 60° 40 40 C/N0 (dB-Hz) C/N0 (dB-Hz) 60° G18 G8 30 30 G11 270° 90° 20 20 PRN8 PRN15 PRN18 PRN11 PRN14 PRN20 10 10 G14 Measured Measured 240° 120° Simulated Simulated 0 0 0 50 100 150 0 50 100 150 0 50 100 150 0 50 100 150 0 50 100 150 0 50 100 150 Time (s) Time (s) 210° 150° 180° (b) Outdoor-indoor-outdoor (c) Outdoor area with strong (a) 3b: yellow, 3c: green transitions blockage Figure 3: Skyplot and C /N0 comparison for GPS L1CA for satellites in Fig. 3a. The receiver is indoor and between two buildings for short time in the middle. The Fig. 3b identifies two modeling issues: Missing objects and vegetation on the east side of the building, and the height of the hall entrance gate. This affects PRNs 15 and 18 on the east side, as well as PRN 20 in Fig. 3c. The PRN 11, high elevation satellite is shortly blocked in the middle at the turning point, resulting in a significant C /N0 drop in the simulated data. This is due to uneven terrain between the two buildings which was not mod- eled. The results show that the 3D-model in its current version agrees with reality in essential aspects covering relevant propagation conditions for GNSS signals. Second, the 5G FR1 rep- resentation is evaluated. The SNR in the Fig. 4 decreases in multipath-rich conditions, for both emulated and simulated data. The underestimation of lower SNRs impacts statistical evaluations in Table 1. The SNRs standard deviation ranges between 5.5 and 8.5 dB for LoS. This is sufficiently accurate since RT channel models show a standard deviation below 8 dB [8]. To compare the positioning performance GNSS and simulated data is used with mostly LoS. Results in Fig. 5a indicate a visible shift between measured and simulated position solu- tion with two-dimensional (2D)-error mean of 18.1 m and a standard deviation of 1.62 m. The atmospheric effects are modeled, resulting in a systematic mean error bias. Finally, the simu- lated 5G FR1 PRS signals are fused using weighted least squares method with measured GNSS signals in Fig. 5b. The measured observations are GPS L1CA and Galileo E1BC pseudoranges, and 5G PRS time of arrival (ToA) values. Here, we use measurements from the TRPs from which the LoSs are present and distinguish two areas: outdoor-indoor-outdoor transitions (left) and outdoor area with strong blockage (right). We compute 5G FR1 position solution for channel data collected: (1) for all TRPs at once and (2) through individual runs for each TRP. The ToA set from (2) is used for sensor data fusion with GNSS. The 5G positioning out- performs both GNSS and hybrid position solution when indoors. In the outdoor case with severe blockage, the 5G signals slightly enhance the GNSS solution because the useful TRPs do not significantly improve the geometrical Dilution of Precision (DOP). 5. Conclusion We have shown that the presented simulation environment is representative for GNSS and 5G FR1 signals under LoS conditions. Additionally, possible modifications for the 3D-model 5G signal 5G signal 50 50 40 40 30 30 20 20 SNR (dB) SNR (dB) 10 10 0 0 -10 -10 -20 -20 -30 Measured -30 Measured Outdoor TRP Indoor TRP Simulated Outdoor TRP Indoor TRP Simulated -40 -40 0 50 100 0 50 100 150 0 50 100 0 50 100 150 Time (s) Time (s) (a) Outdoor-indoor-outdoor transitions (b) Outdoor area with a strong blockage Figure 4: Measured and simulated SNR values for 5G TRPs deployed as in Fig. 2c Table 1 Difference between emulated and simulated 5G SNR values ∆X LoS and ∆X N LoS for LoS and NLoS cases respectively where µ and σ denote sample mean and variance Scenario µ(∆X LoS ) σ(∆X LoS ) µ(∆X N LoS ) σ(∆X N LoS ) outdoor-indoor transition 2.49 dB 8.42 dB −6.63 dB 11.26 dB outdoor with strong blockage 8.41 dB 5.74 dB −2.02 dB 12.66 dB GNSS position solution 40 40 Simulated GNSS Measured 5G-FR1 (1) 35 35 5G-FR1 (2) Simulated w/o bias hybrid 30 30 25 2D-Error (m) 25 North (m) 20 20 15 15 10 10 5 5 0 0 -5 0 50 100 150 200 250 300 350 0 5 10 15 20 25 30 35 40 Snapshot East (m) (a) 2D GNSS position solution comparison (b) Fused 2D position solution Figure 5: 2D position solution improvement are given. Especially, the 5G NLoS case is strongly dependent on the material properties and requires additional study. Brief hybrid positioning results for 5G and GNSS have been shown. With this simulation environment advanced hybrid positioning solutions will be developed. Acknowledgments The research results presented in this paper have been accomplished within a part of the project 5G-Bavaria-Testzentrum which is funded by the Bavarian Ministry of Economic Af- fairs, Regional Development and Energy. We want to thank Mohammad Alawieh, Birendra Ghimire, Ernst Eberlein, and Matthias Overbeck for their tremendous support and valuable discussions. References [1] C.-X. Wang, J. Bian, J. Sun, W. Zhang, M. Zhang, A survey of 5g channel measurements and models, IEEE Communications Surveys & Tutorials 20 (2018) 3142–3168. [2] J. A. del Peral-Rosado, J. A. Lopez-Salcedo, Sunwoo Kim, G. Seco-Granados, Feasibility study of 5G-based localization for assisted driving, in: 2016 International Conference on Localization and GNSS (ICL-GNSS), IEEE, Barcelona, Spain, 2016, pp. 1–6. [3] Ericsson, RAT independent and hybrid NR positioning solutions, 2019. R2-1901531. [4] David Eugen Grimm, Hans-Martin Zogg, Leica Nova MS50 White paper, 2013. [5] J. Seybold, GNSSA-DCP (DUAL CIRCULARLY POLARIZED GNSS ANTENNA), Technical Report, TeleOrbit GmbH, 2020. Available Online: https://teleorbit.eu/en/satnav/. [6] G. M. Tommaso Panicciari, Mohamed Ali Soliman, Evaluation of a gnss receiver perfor- mance in different multipath environments with a novel real-time multipath simulation system, Proceedings of the 30th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2017) (September 2017). [7] F. Mata, F. Grec, M. Azaola, F. Blázquez, A. Fernández, E. Dominguez, G. Cueto- Felgueroso, G. Seco-Granados, J. del Peral-Rosado, E. Staudinger, C. Gentner, M. Kas- parek, C. Backert, D. Barlett, E. Serna, L. Ries, R. Prieto-Cerdeira, Preliminary Field Trials and Simulations Results on Performance of Hybrid Positioning Based on GNSS and 5G Signals, 2020, pp. 387–401. [8] A. W. Mbugua, Y. Chen, L. Raschkowski, L. Thiele, S. Jaeckel, W. Fan, Review on Ray Trac- ing Channel Simulation Accuracy in Sub-6 GHz Outdoor Deployment Scenarios, IEEE Open Journal of Antennas and Propagation 2 (2021) 22–37.