Position Error Analysis of IRNSS Data Using Big Data Analytics
Abstract
The current investigation involves the study and analysis of position error of IRNSS (Indian Regional Navigation Satellite System) receiver system data with respect to a fixed position using big data analytics tool namely, Hadoop. IRNSS is an indigenously developed navigation system by India with constellation of seven satellites. IRNSS data pertaining to week number 918 has been used to carry out the statistical analysis in three different modes of operation. Queries/commands were created in Hadoop to process more than 6 million data (week no. 918 data). The processed data and calculated statistical parameters show that IRNSS navigation system is a promising solution for various navigation related applications. The results obtained highlights that the position error is minimum in Hybrid (GPSL1 + IRNSSS + IRNSSL5) mode of operation with better accuracy when compared with individual frequency mode of operation (L5 & S) of IRNSS receiver system.
Keywords
GNSS IRNSS Satellites Position error Hadoop1 Introduction
The IRNSS satellite signals are available at L5 band (1176.45 × 10^{6} Hz) and S band (2492.028 × 10^{6} Hz) microwave frequencies. The SPS (Standard positioning Service) signal is modulated by 1 × 10^{6} Hz BPSK (Binary Phase shift keying) [6, 7] signal and RS (Restricted Service) signal is modulated with BOC (Binary offset carrier) for high precision and security. To maintain necessary coverage area and signal strength, the signal is being transmitted through phased array antenna.
A Java written open source framework (Apache) big data analytics toolHadoop is used for processing of large datasets in a distributed way across network of computers using robust programming models. Hadoop manages applications with MapReduce algorithm by processing the data on different CPU nodes in parallel. In simple, Hadoop framework allows the users to develop applications that are compatible to run on clusters of computers and also it can achieve complete statistical study for a large quantity of data. In this study, Hadoop frame work is used to analyze the IRNSS data statistically to assess the position error. This paper has been organized with methodology & mathematical modeling in Sect. 2 with simulation results and conclusion in Sects. 3 and 4 respectively.
2 Methodology
2.1 IRNSS Data and HDFS
In general, the IRNSS data consists of files related to 3 solutions namely sol_1 (Mode 1), sol_2 (Mode 2), sol_3 (Mode 3) along with satellite information, PR (pseudo random) log & iono tropo. For week no 918, Sol_1 was assigned as hybrid mode, which is the combination of GPS (L1) + IRNSS (L5 + S), Sol_2 was assigned as IRNSS S Band and Sol_3 was assigned as IRNSS L5 Band. The three solutions (modes) consists of parameters like week no, user time, system time, solution type, iono type, X coordinates, Y coordinates, Z coordinates, latitude, longitude, altitude, geometric dilution of precision (GDOP), position dilution of precision (PDOP), vertical dilution of precision (VDOP), horizontal dilution of precision (HDOP), time dilution of precision (TDOP), position error, number of satellites, velocity information, etc. Satellite information data (file) consists of week no, user time, system time, pseudo random noise (PRN), elevation, azimuth, carrier to noise density ratio (CNO) and X, Y, Z coordinates of all the satellites. PR log data (file) consists of week no, user time, system time, Doppler Frequency (FDOPP) and Iono tropo data (file) consists of week no, user time, system time, iono, tropo information.
2.2 Mathematical Modeling for Statistical Analysis Using Hadoop
The data processed by Hadoop gives the minimum and maximum values of X, Y & Z coordinates, standard deviation, position errors, RMS error and mean position of X, Y & Z. The 3D RMS position is found by taking the mean square error with the help of standard reference values that was calculated by averaging X, Y and Z coordinates for a specific period of time. The ECEF (Earth center earth fixed) values are taken to calculate the mean X, Y, Z positions by using hive queries.
Sample Queries
Data/file  Queries 

Sol_1:  CREATE EXTERNAL TABLE Sol_1(Col_1 datatype, Col_2 datatype ….) 
Sol_2:  CREATE EXTERNAL TABLE Sol_2(Col_1 datatype, Col_2 datatype ….) 
Sol_3:  CREATE EXTERNAL TABLE Sol_3(Col_1 datatype, Col_2 datatype ….) 
Satellite info:  CREATE EXTERNAL TABLE Satellite_info(Col_1 datatype, Col_2 datatype …) 
Iono tropo:  CREATE EXTERNAL TABLE iono_tropo(Col_1 datatype, Col_2 datatype …) 
PR log:  CREATE EXTERNAL TABLE pr_log(Col_1 datatype, Col_2 datatype …) 
Commands:  
Select the data of week no 918:  select * from sol_1 where week no = 918 
Select only the valid data  select * from sol_1 where ux > 0.0 
Bias_X, Bias_Y, Bias_Z:  select ux, uy, uz, round((ux  1345092.73), 2) as bias_x, round((uy  6073169.66), 2) as bias_y, round((uz  1408458.60), 2) as bias_z from sol_1_f 
Mean_X, Mean_Y, Mean_Z:  select avg(bias_x) as mean_x, avg(bias_y) as mean_y, avg(bias_z) as mean_z from sol_1_f_xyz 
Min & Max of X, Y & Z bias:  select min(bias_x) as min_x, max(bias_x) as max_x, min(bias_y) as min_y, max(bias_y) as max_y, min(bias_z) as min_z, max(bias_z) as max_z from sol_1_bias 
Standard deviation:  select stddev_pop(bias_x) as std_x, stddev_pop(bias_y) as std_y, stddev_pop(bias_z) as std_z from sol_1_bias 
ux, uy & uz pos:  select ux, uy, uz from sol_1_f 
Standard deviation of pos:  select stddev_pop(ux) as std_x_pos, stddev_pop(uy) as std_y_pos, stddev_pop(uz) as std_z_pos from sol_11 
Mean_X, Mean_Y, Mean_Zpos:  select avg(ux) as mean_x_pos, avg(uy) as mean_y_pos, avg(uz) as mean_z_pos from sol_11 
Sum of bias_X, bias_Y, bias_Z:  select sum(bias_x) as sum_x, sum(bias_y) as sum_y, sum(bias_z) as sum_z from sol_1_bias 
3 Results and Discussion
Statistical analysis of GPS & IRNSS data using Hadoop framework for week no. 918
Parameters  Hybrid: GPSL1 + IRNSSS + IRNSSL5 (m)  IRNSSS (m)  IRNSSL5 (m) 

3D RMS position error  3.24727  3.23492  3.235524 
Max X error  46.23  2.63  4.88 
Min X error  −16.91  −8.63  −9 
Max Y error  87.08  39.73  47.83 
Min Y error  10.95  20.72  19.81 
Max Z error  6.76  6.05  7.72 
Min Z error  −20.01  −103.35  −102.7 
Mean X error  4.97  −5.06  −7.28 
Mean Y error  0.0029  7.54  −4.02 
Mean Z error  −8.12  −5.55  −8.59 
SD X error  0.35  0.57  0.84 
SD Y error  0.96  1.41  1.55 
SD Z error  0.57  1.42  2.02 
Mean X position  1345097.70  1345087.66  1345085.44 
Mean Y position  6073169.66  6073177.20  6073165.63 
Mean Z position  1408450.47  1408453.04  1408450.00 
Reference X position  1345092.73  
Reference Y position  6073169.66  
Reference Z position  1408458.60 

Rows 3, 5 & 7 shows that IRNSSS band is giving better results in terms of maximum position error (X, Y & Z) compared to Hybrid and IRNSSL5 band.

From rows 4 & 6, it is observed that hybrid mode is better in giving minimum position error for X & Y coordinates compared to IRNSSS & L5 band. Row 8 shows that IRNSSS band gives better in terms of minimum position error for Z coordinates in comparison with Hybrid and IRNSSL5 band.

From row 12, 13 & 14, it is seen that the standard deviation (X, Y & Z error) with respect to reference value is minimum in hybrid mode in comparison with IRNSSS & L5 band. The second best solution is IRNSSS band.

From rows 9, 10 & 11, it is observed that IRNSSL5 band is giving minimum average position error when compared to Hybrid and IRNSSS band.

From row 15 & 18, it is seen than the difference between mean X position and reference X position is minimum in hybrid solution compared to IRNSSS & L5 band.

From row 16 & 19, it is seen than the difference between mean Y position and reference Y position is minimum in hybrid solution compared to IRNSSS & L5 band.

From row 17 & 20, it is seen than the difference between mean Z position and reference Z position is minimum in IRNSSS band solution compared to hybrid & IRNSSL5 band.
The proposed study shows that hybrid mode of operation is giving the best position information with minimum position error and maximum accuracy than the individual L5 or S band of IRNSS receiver system. The second best solution is given by IRNSSL5 band.
4 Conclusion
This study is an attempt to utilize open source tools like Hadoop to carry out the statistical analysis of position error from the IRNSS data. The Data is collected in three different modes like Hybrid (GPS + IRNSS (L5 + S)), IRNSS(S) and IRNSS (L5) and processed individually to measure the effectiveness of IRNSS. This study cum analysis shows that, there is an improvement in accuracy and visibility with respect to hybrid mode of operation with IRNSS L5 band giving the second best solution. The position error analysis computed using Hadoop for IRNSS receiver, is essential for understanding how IRNSS works, and for knowing what degree of errors could be anticipated. This helps in making corrections for receiver clock errors and other effects. This kind of analysis could be carried out for any GNSS receivers. This paper provides an overview for researchers in GNSS field to choose the appropriate solution depending on the applications.
Notes
Acknowledgment
The authors would like to thank SAC ISROAhmadabad, Sri Sringeri Sharada PeethamSringeri and Jyothy Institute of TechnologyBangalore for providing an opportunity to work with IRNSS receiver system.
References
 1.Montenbruck, O., Steigenberger, P.: IRNSS orbit determination and broadcast ephemeris assessment. In: ION International Technical Meeting (ITM), Dana Point, CA, pp. 26–28 (2015)Google Scholar
 2.Raghu, N., Manjunatha, K.N., Kiran, B.: Determination and preliminary analysis of position accuracy on IRNSS satellites. In: International Conference on Communication and Signal Processing, pp. 6–8. IEEE Press, India (2016). https://doi.org/10.1109/iccsp.2016.7754248
 3.Khatri, R.R., Chauhan, S.: Indian regional navigation satellite system. Int. J. Innov. Res. Technol. 2, 380–384 (2016)Google Scholar
 4.Indian Space Research Organization. https://www.isro.gov.in/irnssprogramme
 5.Raghu, N., Kiran, B., Manjunatha, K.N.: Tracking of IRNSS, GPS and hybrid satellites by using IRNSS receiver in STK simulation. In: International Conference on Communication and Signal Processing, India, pp. 6–8 (2016). https://doi.org/10.1109/iccsp.2016.7754276
 6.Guangya, X.W., Zhiqiang, S., Xiaofeng, L., Yang, H.J.: Research on the 3D visualization of information operations based on STK. In: International Conference on Audio Language and Image Processing, pp. 921–925. IEEE Press, China (2010). https://doi.org/10.1109/icalip.2010.5685180
 7.Fan, S., Zhao, L., Xiao, W., Li, Z.: Performance analysis and simulation of iridium navigation satellite based on STK. In: 2nd International Workshop on Earth Observation and Remote Sensing Applications, China, pp. 291–295 (2012). https://doi.org/10.1109/eorsa.2012.6261185