Site-specific Multipath Characteristic of GPS ISKANDAR Network

Transcription

1 Site-specific Multipath Characteristic of GPS ISKANDAR Network NOOR SURYATI M. S. & MUSA, T. A. UTM-GNSS & Geodynamics Research Group, Faculty of Geoinformation Science & Engineering, Universiti Teknologi Malaysia (UTM), Skudai, Johor, MALAYSIA. Tel.: Fax: Abstract The rise of a new economic region called ISKANDAR Malaysia has great demand for high precision positioning system. By establishing ISKANDARnet, a Network-RTK positioning system in this region, precise positioning can be achieved even in real-time. The idea of this system is to model errors, especially distance-dependent errors present within a network by deploying multiple reference stations. However, there are errors that cannot totally be modelled such as multipath effects. Since the multipath effect depends on the site environment, the suitability of site for reference station installation has to be determined. Hence, this study intends to interpret multipath effects at reference stations during the establishment of ISKANDARnet. The multipath effects at each station are evaluated by numerical calculation and graphical analysis. The suitability of the location of stations will then be judged. This crucial step has yielded valuable experience in ensuring good data quality. It is an essential issue that must be considered to enable the network-based positioning works effectively, especially for development of ISKANDAR Malaysia. Keyword: Network-RTK, CORS, multipath 1.0 INTRODUCTION The ideal development region of Southern Peninsular Malaysia, called ISKANDAR Malaysia has lead to increased substantial activities particularly in construction and infrastructure. These activities normally include high precision positioning that requires reliable positioning technique. An approach of network-based positioning currently preferred to be implemented in order to obtain centimetrelevel positioning accuracy in real-time over large area. This system typically named as Network- RTK, is a carrier phase-based positioning technique which is combining and interpolating measurements from multiple reference stations and thus, generates network corrections to be applied by network users in extended baseline. Over the last decade, there are many countries have implemented the Network-RTK positioning system such as in Australia (SydNet), Germany (SAPOS), Denmark (REFDK), Hong Kong (SatRef), Japan (GEONET) and, in Malaysia, known as MyRTKnet. In ISKANDAR Malaysia region, a Network-RTK system called ISKANDARnet is being established by Universiti Teknologi Malaysia - GNSS and Geodynamics (UTM-G&G) research group. Currently, ISKANDARnet has deployed a permanent Continuously Operating Reference Station (CORS) and is going to proceed for other CORS installation. The crucial task during CORS installation is to judge site selection. It is essential in order to ensure the quality of data measurements at every CORS. Good data quality supplies sufficient network corrections to user (Suryati et al., 2008). Note that network correction implies to model distancedependent errors (i.e. atmospheric and orbital error). However, there are errors cannot completely be modelled especially the multipath effects. This multipath error is varies and is influenced by 1

2 surrounding environments at the Global Positioning System (GPS) station location. Basically, multipath is the reception of the GPS signal via multiple paths rather than from a direct line of sight (Van Sickle, 2008). Consequently, this phenomenon will introduce errors in GPS measurement which is difficult to eliminate unless reducing it by suitable site selection. In this study, insight of GPS site selection is being understood as initial steps of establishing the ISKANDARnet. Outlines of data quality analysis are conducted in two cases which are for ISKANDARnet1 and ISKANDARnet2 stations. Evaluation and interpretation of multipath effects at both stations are realised by values of multipath on L1/L2 and polar plots. 2.0 THE MULTIPATH EFFECTS The multipath effect is mainly caused by reflection of satellite signals on a reflecting surface objects. For GPS signals, this effect appears in the situation such as near the buildings or other elevations, metallic structures, or water surfaces. The satellite signal does not travel directly to the antenna but hits the nearby object first and is reflected into the antenna creating a false measurement. The measurement errors also depend upon the satellite elevation angle. As a satellite gets lower in the sky, the received signal power decreases and multipath increases (Misra and Enge, 2001). Figure 1 shows how multipath occur and affect the signals. Figure 1: Multipath Effects (via There is no general model of the multipath effects because of the time- and location-dependent geometric situation (Hofmann-Wellenhof et al., 2008). However, the effects can be estimated by using a combination of frequency f 1 and f 2 code and carrier. Its normally can be computed into multipath on L1 (mp1) and L2 (mp2) as equation (1) and (2) respectively (Estey and Meertens, 1999). mp1 2 2 P 1 1 L1 L2 (1) 1 1 mp2 2 2 P 2 L1 1 L2 (2) 1 1 Where P is the observation of pseudorange and L is the observation of the carrier phase, α = (f 1 /f 2 ) 2, and f 1, f 2 denote the frequencies of L1 and L2, respectively. 2

3 Besides, the multipath effects may be reduced by using antenna choke rings, wideband antenna, antenna ground planes and digital filtering. There is also a technique based on an adaptive filter using the least-mean-square algorithm which is proposed by Ge et al. (2000). However, the simplest way is to avoid sites where it could be a multipath problem. 3.0 DATA QUALITY ANALYSIS Data quality particularly involving multipath effect is important to be analysed for suitability of station location. The analysis can be done by computing Root Mean Square of multipath on L1 (RMS mp1) and on L2 (RMS mp2). The Translation, Editing, and Quality Check (TEQC) program was executed to get the RMS mp1 and RMS mp2 values. The TEQC also provides output files such as signal-to-noise ratios (SNR) on L1 and L2, satellite elevation and azimuth, ionospheric, and derivative of ionospheric delay. The multipath files containing L1 or L2 multipath, satellite elevation and azimuth can be used to further examine the multipath effects with Teqcspec. Teqcspec is a MATLAB code for creating colorized polar plots (Ogaja and Hedfors, 2007), which enable interpretation of multipath effects at the site. In this study, multipath effects at ISKANDARnet1 and ISKANDARnet2 stations are analysed. Case 1: Analysis of multipath effects at ISKANDARnet1 a) RMS mp1 and mp2 The multipath effects at ISKANDARnet1 station are analysed by taking one month RINEX data; 1 st to 30 th April 2009 (i.e. Day of Year 91 to 120). The observations are logged into 15 seconds interval during 24 hours and satellite elevation cut-off angle was set at 15 degrees. The results of RMS mp1 and mp2 which computed by TEQC are shown in Figure 2 and 3. Figure 2: Multipath effects on L1 at ISKANDARnet1 Station. 3

4 Figure 3: Multipath effects on L2 of ISKANDARnet1 Station. Figure 2 indicates the RMS mp1 for ISKANDARnet1 station are dominated by the magnitude of 0.06 m. There is only five days during the observation week gave the result of 0.07 m RMS mp1. Meanwhile, Figure 3 indicates the RMS mp2 values range from 0.12 m to 0.15 m which is higher than mp1. The high value in mp2 seems to show that there are multipath errors respect to the linear combinations of the carrier phase and pseudorange observations as shown in equation (1) and (2). According to Bruyninx et al. (2003), 75 percent of the European reference stations namely EUREF Permanent Network (EPN) has mp1 values below 0.57 m and mp2 values below 1 m. Furthermore, 50 percent of the International GNSS Service (IGS) stations around the world have RMS of mp1 under 0.4 m, and 75 percent have less than 0.5 m. Meanwhile, the RMS mp2 value for 50 percent of IGS stations are less than 0.6 m and 75 percent are less than 0.75 m. It is signify that the multipath effects at ISKANDARnet1 station is in the range of acceptable. b) Polar Plots with Multipath, Satellite Elevation and Azimuth The polar plots of ISKANDARnet1 station have been created by taking 24 hours observation data recorded on 26 th April The satellite elevation cut-off angle was set at 15 degrees. The multipath effects on L1 for every satellite is shown in Figure 4 with the colour scale bar indicate in metre. Figure 5 and 6 illustrate elevation and azimuth of satellites during the observation in degrees. Note that dashed black lines in these figures show time windows as related to the polar plots. 4

5 Figure 4: Multipath Effects on L1 at ISKANDARnet1. Figure 5: Satellite Elevation. 5

6 Figure 6: Satellite Azimuth. The polar plots were organised into four time windows based on Coordinated Universal Time (UTC); 00:00:15 06:00:15, 06:00:15 12:00:15, 12:00:15 18:00:15, and 18:00:15 23:59:15. Figures 7 to 10 show each of time windows of 6 hours. The main reason of this separated four time windows is to aid the analysis and realisation of multipath effects at the site. The angle of azimuth (0 о to 360 о ) is corresponding to the satellite azimuth and the black dashed circle indicates the satellite elevation (0 о to 90 о ) from horizontal. Whereas the coloured lines show the path of each satellite during the observation which satellite numbers are specified at the end of the satellite track. 90 о 90 о 6 Figure 7: Polar Plot of First Time Window Figure 8: Polar Plot of Second Time Window

7 90 о 90 о Figure 9: Polar Plot of Third Time Window Figure 10: Polar Plot of Fourth Time Window The red circles in Figures 7 to 10 indicate the suspected multipath effects. In this case, high multipath effects appeared in south-east direction and at satellite elevation of 15 о up to 50 о. Specifically, all polar plots (Figures 7-10) gave a repeated multipath effects when reach about 130 о azimuth and 35 о elevation. It is significant that multipath effects also occur at this similar area even satellite elevation up to about 50 о (see satellite S30 and S18 in Figure 8). The reason of this phenomenon is revealed in Figure 11 as it is clearly shown that the mosque is situated at the south-east direction of ISKANDARnet1. Although the distance between ISKANDARnet1 station and the mosque is about 200 m, the tall dome and tower at the mosque most probably contribute multipath effects. The mosque s tower can be seen in Figure 12 (a). N W E S Figure 11: Map of ISKANDARnet1 Environment. 7

8 GPS Antenna Mosque s tower FAB Building C03, FKSG (a) Figure 12: Site Environment of ISKANDARnet1. (b) Strong multipath effects also often occurred in azimuth of 180 о to 190 о especially when satellites rise at about 30 о from horizon. By referring to Figure 11, there is building surrounded by this azimuth. A site inspection reveals that the Faculty of Built Environment (FAB) building is situated only about 20 m from the GPS antenna in that azimuth. Figure 12 (a) shows the site environment with the FAB building is almost equal height as the C03, FKSG building, where the GPS antenna is mounted. Figure 10 shows there is also high multipath effect in the west direction. From the site inspection, there is a communication tower (see Figure 12 b) located at the west of ISKANDARnet1. Other direction of ISKANDARnet1 has only small amount of multipath effects typically caused by low satellite elevation. Case 2: Analysis of multipath effects at ISKANDARnet2 a) RMS mp1 and mp2 ISKANDARnet2 station has been temporarily established at Port of Tanjung Pelepas (PTP) in order to study the suitability of site. The testing was held on 3 rd to 6 th April, 2009 on rooftop of PTP building, Gelang Patah, Johor. All GPS data were recorded for every 15 seconds interval at 15 degrees of satellite elevation cut-off angle. The analysis of multipath effects was conducted by calculating the RMS of mp1 and mp2 via TEQC. The results of RMS mp1 and mp2 are shown in Figure 13 and 14 respectively. Figure 13: Multipath effects on L1 at ISKANDARnet2 Station. 8

9 Figure 14: Multipath effects on L1 at ISKANDARnet2 Station. Based on Figure 13, the RMS mp1 values for ISKANDARnet2 is vary from 0.15 m to 0.17 m in four days. Meanwhile, Figure 14 shows the RMS mp2 is 0.22 m to 0.25 m. However, the RMS mp1 and mp2 values are still in the range of acceptable by referring to quality status of other CORS station and IGS station. Figure 13 and 14 also show that the pattern of RMS mp1 and mp2 are almost the same. Perhaps, the same factors are affecting the L1 and L2 signal. Further investigation on source of the multipath errors can be done by reviewing polar plot and site inspection. b) Polar Plots with Multipath, Satellite Elevation and Azimuth In this case, ISKANDARnet2 station uses 24 hours data on 5 th April 2009, to create polar plots for interpretation of multipath effects. The interpretation of multipath orientation can be further by considering the supplement graphs such as multipath on L1, satellite elevation and satellite azimuth as shown in Figures 15, 16 and 17, respectively. The procedure and feature of multipath interpretation is identical in Case 1. Figure 15: Multipath Effects on L1 at ISKANDARnet2. 9

10 Figure 16: Satellite Elevation. Figure 17: Satellite Azimuth. 10

11 90 о 90 о Figure 18: Polar Plot of First Time Window Figure 19: Polar Plot of Second Time Window 90 о 90 Figure 20: Polar Plot of Third Time Window Figure 21: Polar Plot of Fourth Time Window The polar plots of ISKANDARnet2 (Figures 18 to 21) show that the azimuth range of 180 о to 210 о are prone to have high multipath effects. This effect is caused by satellites at elevation of about 40 о. Another location that strongly affected by multipath is specified at azimuth 290 о and elevation about 35 о can be seen in all figures. Additionally, there are also multipath effects in the azimuth of 120 о to 180 о, especially when satellites reach at about 15 о to 30 о elevation. The repetition of multipath effects can be further explained by site inspection. The strong multipath in azimuth range of 180 о to 210 о is most probably because of the communication antenna nearby as can be illustrated in Figure 22 (a). The antenna is purposely to aid vessel, thus place at the South direction facing the port. Besides that, the occurrence of multipath effects at 11

12 azimuth 290 о is caused by the tall object nearby as shown in Figure 22(b). Nevertheless, the multipath effects can be reduced by designing a taller standing bracket for GPS antenna. GPS Antenna Other antenna Other object (a) (b) Figure 22: GPS antenna at the rooftop of PTP building with the other (a) communication antenna and (b) objects nearby. 6.0 CONCLUDING REMARKS & FUTURE WORK In this paper, multipath effect was checked at every reference station as part of data quality check and site selection. Multipath effect at ISKANDARnet2 is higher than ISKANDARnet1, however multipath effects for the both stations are acceptable. Polar plots aid the multipath interpretation and thus reveal the orientation of multipath effects at the site. Furthermore, the site inspections convince the interpretation that multipath was affected by nearby and taller objects. Establishment of another two CORS stations for ISKANDARnet will be carried out very soon. Then, the same procedure for multipath analysis will be applied. It is expected that by minimizing the effect of multipath during the CORS station installation could support high precision applications within the metro-area of ISKANDAR Malaysia. 12

13 REFERENCES Bruyninx, C., Carpentier, G., and Roosbeek, F. (2003).Today s EPN and Its Network Coordination. EUREF Symposium, 4-6 June, Toledo, Spain. 13 (33), EUREF Publication. Estey, L., H., and Merteens, C., M. (1999). TEQC: The Multi-Purpose Toolkit for GPS/GLONASS Data. GPS Solutions. 3 (1), John Wiley & Sons, Inc. Ge, L., Han, S., and Rizos, C. (2000). Multipath Mitigation of Continuous GPS Moisuromonts Using an Adaptive Filter. GPS Solutions. 4 (2), John Wiley & Sons, Inc. Hofmann-Wellenhof, B., Lichtenegger, H. and Wasle, E. (2008). GNSS Global Navigation Satellite Systems GPS, GLONASS, Galileo and more. Austria: SpringerWienNewYork. Misra, P. and Enge, P. (2001). Global Positioning System Signals, Measurements, and Performance. USA: Ganga-Jamuna Press. Ogaja, C. and Hedfors, J. (2007). TEQC Multipath Metrics in MATLAB. GPS Solutions. 11, , Springer-Verlag. Suryati, M. S., Musa, T. A., Ses, S. (2008). Impact of Poor Data Correction to Network- RTK User s. International Symposium on Geoinformation, Octob er. Kuala Lumpur, Malaysia. Van Sickle. J. (2008). GPS for Land Surveyors, Third Edition. Boca Raton, Landon: CRS Press. Woessnes, M. (2005). Sources of Errors in GPS. Accessed via Accessed: 10 March