Home Articles A study on the utility of combined use of GPS and GLONASS...

A study on the utility of combined use of GPS and GLONASS constellations in India

P. Banerjee


P. Banerjee
National Physical
Laboratory,New Delhi
[email protected]

Anindya Bose
Physics Department Burdwan
University Burdwan

A. Dasgupta
Institute of Radio Physics & Electronics Calcutta University
Calcutta

Abstract
GPS has been operational with its full constellation since early 90’s. The GLONASS constellation is also now operational. The effort of integrating two constellations resulted in the production of receiver, which may use both constellations in combination. Utilising this type of receiver GPS and GLONASS signals were monitored for one calendar year in different places in India simultaneously. The aim of this monitoring was to study aspects of visibility of satellites and the limits of accuracy. It has been observed that on the average 95% of time, 8 or more GPS satellites are available anywhere in India. Number of satellites in GLONASS constellation has been found to be gradually depleting form 16 to 7 during the course of this study. So 3D solution thus was hardly possible by the use of only GLONASS satellites. GLONASS alone, however, may be used for timing application as a single satellite can also give timing solution at a predetermined location. However, appreciable improvement in PDOP was observed in the combined mode. Before the withdrawal of GPS SA significant improvement of position accuracy could be observed in the combined mode. After the removal of GPS-SA, the accuracy of the combined mode and that of only GPS mode have been found to be of the same order. It apparently does not reflect any advantage. But it indirectly confirms that both, GPS and GLONASS systems have similar accuracy limits of accuracy and also confirms that interoperability of two systems has optimally been achieved. These studies reveal that the use of GLONASS along with GPS will always be beneficial to the varying degree depending on different application and circumstances. This paper elaborates these observations.

Introduction
Both GPS and GLONASS will offer, independently, precise location and time transfer continuously anywhere in the world and, indeed, in space itself. The reliability and availability of GPS in Indian subcontinent have already been exhaustively studied and presented in the report of SRAGI project. It has been felt that GPS alone fall short of meeting the requirements of a global sole means, or stand-alone, navigation system. GLONASS alone also cannot serve the same purpose. The performance capabilities of GLONASS are substantially similar to those of GPS as claimed by the respective authorities. But GPS has achieved its operation status in full, while the prospects of GLONASS are less clear. The two systems taken together, however, should offer amply redundant measurements to all users, and seem capable of meeting these requirements. To study these aspects, GPS and GLONASS signals were monitored for one year in different places in India. The number of satellites with good signal strength, the geometry of the avalable satellites and the corresponding achievable accuracy were recorded during this study. This paper attempts to examine the performance capabilities of GLONASS and compare them with those of GPS based on monitored data Constellation Status of GLONASS and GPS It is necessary to have a prior knowledge of constellation status of both GPS and GLONASS in order to have the comprehensive conclusion on the availability of the satellites. The GLONASS satellite was launched in 1982. 24-satellite was available for the first time on 18 January 1996. This status, however, was relatively short-lived and the constellation has gradually fallen into decline, as failed satellites have not been replaced. Since December 1995 there was no launch till December1998 when three new satellites were launched. For replenishment after a gap of two years three satellites were launched in December 2000 but only one of them has been successfully operational. The status of the constellation during the period of Dec. 1998 to Feb 2001 has been tabulated in Table. 1.

GPS constellation has been full with 24 satellites or more during the entire period of this study. It may be noted that there were 29 satellites operational. Out of these, 6 satellites were launched in the year 1900 or before.

Availability of GLONASS and GPS Satellites in India
The monitoring of GPS and GLONASS satellites has been carried out at three sites (New Delhi, Calcutta and Bangalore) independently and the recorded data had been analysed. The number of satellites in GLONASS constellation varied widely during the period of study. So the availability of GLONASS should be referenced to its constellation status. The type of receivers used in this campaign may be operated in any one of three modes namely in GPS only mode, GLONASS only mode and GPS and GLONASS combined mode. Constellation of GPS was full with 24 or more satellites during the entire period of observation. So GPS constellation status may be assumed to be uniform during the period of this study. When the receiver operates in GPS+GLONASS mode, each sample of observation displays the elevation and azimuth of all
 Table 1: Status of the constellation.

Update Date Satellites Total No. of Satellites
December 1998 01, 03, 04, 06, 09, 10, 11, 13, 14, 16, 18, 20, 22, 23 14
March 1999 01, 03, 04, 06, 07, 08, 09, 10, 11, 13, 14, 16, 20, 22, 23 15
June 1999 01, 03, 04, 06, 09, 10, 11, 13, 14, 15, 16, 18, 20, 22, 23 15
September 1999 01, 06, 09, 10, 11, 13, 15, 16, 18, 22, 23 11
December 1999 01, 03, 04, 09, 10, 11, 13, 14, 15, 16, 18, 20, 22, 23 14
March 2000 01, 03, 04, 09, 10, 11, 13, 14, 15, 16, 18, 20, 22, 23 14
June 2000 01, 03, 04, 09, 10, 11, 13, 14, 15, 16, 18, 20, 22, 23 14
September 2000 01, 03, 04, 10, 11, 13, 14, 15, 18 9
December 2000 01, 03, 04, 10, 11, 13, 15, 17, 18 8
23.01.2001 1, 7, 8, 11, 13, 15, 17 7

GPS and GLONASS satellites being used for finding position solution. These satellites are counted. These numbers have been analysed in a very comprehensive manner. As the availability of GPS and GLONASS signal has been extensively studied at three sites well scattered over India, the observation may be assumed to general observation for Indian subcontinent. The observation does not project any diurnal or seasonal nature variability. Also it has been seen that the availability of signal from both the constellation is almost similar in nature with insignificant variability over whole of India.


Fig.1: Probability of a certain number or more GPS satellites on different days.
Some of observations on GPS availability is shown in Fig.1. On the average 95% of time, 8 or more GPS satellites are available any where in India. Even maximum number of 13 satellites was available in Calcutta in Bangalore but New Delhi never received more than 12 satellites.

It is very difficult to make general conclusion on the availability of GLONASS satellites as number of satellites in GLONASS constellation has been found to be gradually depleting during the course of this study. This may be attributed to the facts that the satellites were not surviving its full life time and also the scheduled replenishment was not carried out possibly due to economic and political reasons. During the period of study the number of satellites in the constellation varied widely from 16 to 7. For different constellation status the visibility of GLONASS is shown in Fig.2. For the best status of 16 satellites, 95% of time 3 or more satellites were available whereas only 1 or more satellites were available for worst status of 7 satellites. In other words, GLONASS constellation was never so good that one may have 3D solution out of only GLONASS constellation. However, GLONASS alone may be used for timing application.


Fig.2: Probability of a certain number or more GLONASS satellites with varying status of the constellation.
 Geometrical Configuration of Available Satellites
It has been noted that sufficient number of GPS satellites or GPS+GLONASS satellites is always available at all the monitoring sites to get the navigation solution. But to achieve the desired accuracy, it is not only necessary to track sufficient number of satellites but also it is equally important that the geometrical configuration of the tracked satellites is also favourable. The effect of geometrical configuration is numerically expressed by the factor Position Dilution of Precision (PDOP). The following subsection presents the detailed analysis of PDOP value of monitored data and also probability of occurrence of different values of PDOP. It is obvious the PDOP is dependent on the position of the receiver and changes with time due to relative motion of the satellites. All modern receivers find out PDOP for all combination of available satellites and select the set of satellites with minimum PDOP for navigation fix. The variation of PDOP for each day of observation has been studied. Percentage of occurrence of PDOP has been found out by just counting the number of samples that corresponds to PDOP values within a particular range.

Fig.3: Probability that PDOP remains within a certain value.
From the availability study it is clear that present status of GLONASS constellation does not encourage using these satellites alone for navigation applications. So attention was given to explore the use of GLONASS in combination of GPS. So the impact on PDOP by the presence of GLONASS satellites along with GPS satellites was emphasized. As expected, appreciable improvement in PDOP in the combined mode by the supplement of GLONASS satellites was observed (e.g. PDOP of 2 in GPS only mode has been improved to 1.6 for combined mode) as shown in Fig.3. But it is of interest to note that the improvement of PDOP was hardly effected by depletion of GLONASS satellites. This apparently implies that the improvement of PDOP may be optimally achieved by addition of only one or two more satellite suitably placed. Further the addition of satellites with the improvement of GLONASS constellation status, may not definitely make any further perceptible impact on PDOP. But it should be noted that the addition of GLONASS satellites surely improves the probability of having at least one or two more GLONASS satellites in favourable position and thereby showing more confidence in improvement of PDOP.

Study on the Position Accuracy
It is necessary to know the precise coordinates of the location of the antenna to study the position accuracy. So the coordinates of all the three monitoring sites were known a-priori fairly accurately. With the help of predetermined position coordinates, the error in the latitude, longitude and height corresponding to each observation may be found out. But to find out the 3D error the following relation have to be used.


where Lt= Nominal value of latitude
Dh = Departure of the observed height from the known height in meters
DLn= Departure of the observed longitude (in minute) from the known longitude, and
DLt= Departure of the observed latitude (in minute) from the known latitude.
The above formulation assumes that 1 minute of arc of the earth curvature is equivalent to 1852 meters.

In Presence of SA
Since May 1, 2000, GPS Selective Availability (SA) has been withdrawn by USA. Observations taken before May 1, 2000 have been studied in a perspective that assumes the presence of SA. SA had been implemented in Block II satellites and had been effective since April 1990 intermittently at various level of accuracy. According to US DoD, the introduction of SA would provide a horizontal accuracy within 100 meters and vertical accuracy within 159 meters for standard positioning service (SPS) with C/A code. The time output accuracy would deteriorate to 350 nanoseconds. One of the primary objectives of this effort was to study the optimal use of GLONASS constellation in combination of GPS to reduce the effect of SA in position accuracy.

In one special campaign, the receiver has been made to operate in three different modes (GPS only, GLONASS only and GPS+GLONASS mode respectively) for 10 minutes each sequentially and the operation were repeated for few hours. In each slot the position solutions corresponding to the each sample of observation has been recorded. Errors in latitude and longitude have been shown for different modes of operations. Numbers at the top in each slot indicate the average number of satellites that participated in the observation. In the mixed mode the first number corresponds to GPS and second number is for GLONASS. These numbers are very useful for proper assessment in right perspective. In GPS only mode the wide fluctuation in error is quite significant reflecting the effect of SA. The diminishing effect of SA in GLONASS only mode and GPS+GLONASS mode has been obviated by smooth variation of the errors. One may argue that due to total absence of SA in GLONASS, the GLONASS only would exhibit best performance. Due to insufficient number of GLONASS satellites, the optimum configuration of satellite geometry would not be possible leading to higher PDOP in GLONASS only mode. But in combined mode, good number of GPS satellites supplemented by fewer GLONASS satellites would allow the selection of satellites with better PDOP. Contribution of GLONASS satellites by higher number helps in more efficient countering of the telling effect of SA.
The effect of SA may to some extent be lessened through averaging over time. The position error has been analysed by averaging them over different length of time. These analyses were compared for different modes of operation. It has been observed that in GPS only mode, error in the time-averaged position data reduces with the increase of averaging time. No obvious improvement of position error could be seen with further increase of averaging time. Similar exercise was done with data recorded in GPS+GLONASS mode. It has been noted that hardly any improvement in error in positioning could be observed by averaging. There is no change in the pattern of position error with the increase of averaging time in this mode. This indicates that the effect of SA had been reduced sufficiently in each sample of observation in GPS+GLONASS mode of operation. So averaging process would not improve further the accuracy significantly.
Table.2: GPS Error budget with and without SA

Error Source With SA Without SA
Selective Availability 24.0 0.0
Ionospheric Delay 7.0 7.0
Tropospheric Delay 2.0 0.2
Clock and Ephemeris Error 2.3 2.3
Receiver Noise 0.6 0.6
Multipath 1.5 1.5
User Equivalent range Error 25.0 7.5
HDOP Error 1.5 1.5
Total stand Alone 75.0 11.5
Horizontal Accuracy    

The status of position error has also been analysed in grouping the error of certain range. For a certain period of observation, say for one complete day, the number of samples found with an error of 5mts or less, with an error of 10mts or less and so on, is found out and these numbers are normalised against the total number of samples. These help in indicating the probability of the error to be within the respective range. It may be noted that probability of error to be within 5mts or 10mts is much higher in mixed mode than that in GPS only mode. On some particular day, the probability of position error remaining within 5mts or less, was as high as 77% in mixed mode and whereas it was only 10% in GPS only mode. It is clear that more than 90% of time the position error remains within 10mts in mixed mode but in GPS only mode the position error goes as high as 30mts.

In Absence of SA
The intentional degradation of the GPS Standard Positioning Service (SPS) known as Selective Availability (SA), has been discontinued from may 1, 2000. Since SA was discontinued, the accuracy of GPS has been expected to considerably improve. So the position error has been reviewed with altogether different perspective. GLONASS promises 60 meters of horizontal accuracy with C/A code. At this point it quite relevant to look at the comparative error budget for GPS positioning and timing with SA and without SA as shown in Table 2

Let us look at the observation made after May 1, 2000. If one refers to Table 2, significant improvement of position accuracy is expected after removal of SA. A sharp improvement in accuracy with scatter reducing from 90mts to 18mts is so evident in these figures. The considerable improvement in position accuracy for GPS only mode was anticipated. In view of this it is very important to properly compare the performance of GPS only mode and GOS+GLONASS mode. To achieve this objective the receiver has been operated 10 minutes in GPS only mode and next 10 minute was operated in mixed mode.

flipping of mode of operation after every 10 minutes of observation was continued for the whole day. The data thus collected were separated for the respective mode of operation and compared. Accuracy was found to be equal or marginally less in the combined mode in comparison to only GPS mode. It shows that the probability of a certain position accuracy is same or slightly greater in only GPS mode than that of the mixed mode of operation.

It is well established that the introduction of SA caused a considerable deterioration of position accuracy in GPS. Through this study it has also been amply demonstrated that the accuracy denied by the GPS-SA may be considerably compensated when the receiver is operated in GPS+GLONASS mode. This observation not only proves that the GLONASS also has better accuracy capability than that of GPS-SA showing the improvement in accuracy in combined mode but also the coordination of two systems could be successfully achieved.

After the removal of GPS-SA, the accuracy of the combined mode and that of only GPS mode have been found to be comparable. It indirectly confirms that both GPS and GLONASS systems have similar accuracy limits and interoperability of two systems has been to a great extent achieved. The slight decrease in accuracy on combined mode may be attributed to the non-near perfect conversion matrix from PZ-90 to WGS-84.

Concluding Remarks
The usefulness of combined use of GPS and GLONASS in India has been substantiated by careful analysis of observations. For the paucity of space, few samples of the data have been illustrated. But general conclusions that have been drawn are based on ensemble of observations made at different places for almost one year.

Acknowledgement
Authors are thankful to MIT and DRDO for supporting this work. Authors are also thankful to CABS, Bangalore for experimental supports and co-operations.

Reference

  • P.Banerjee and Anindya Bose, ” Study on the Reliability and Availability of GPS Signal in India (SRAGI)”, March 1997, NPL, New Delhi.