Interpolation Method Research and Precision Analysis of GPS Satellite Position

Jianmin WANG, Yabo LI, Huizhong ZHU, Tianming MA

Journal of Systems Science and Information ›› 2018, Vol. 6 ›› Issue (3) : 277-288.

PDF(182 KB)
PDF(182 KB)
Journal of Systems Science and Information ›› 2018, Vol. 6 ›› Issue (3) : 277-288. DOI: 10.21078/JSSI-2018-277-12
 

Interpolation Method Research and Precision Analysis of GPS Satellite Position

Author information +
History +

Abstract

According to the precise ephemeris has only provided satellite position that is discrete not any time, so propose that make use of interpolation method to calculate satellite position at any time. The essay take advantage of IGS precise ephemeris data to calculate satellite position at some time by using Lagrange interpolation, Newton interpolation, Hermite interpolation, Cubic spline interpolation method, Chebyshev fitting method respectively, which has a deeply analysis in the precision of five interpolations. The results show that the precision of Cubic spline interpolation method is the worst, the precision of Chebyshev fitting is better than Hermite interpolation method. Lagrange interpolation and Newton interpolation are better than other methods in precision. Newton interpolation method has the advantages of high speed and high precision. Therefore, Newton interpolation method has a certain scientific significance and practical value to get the position of the satellite quickly and accurately.

Key words

precise ephemeris / satellite position / lagrange interpolation / Newton interpolation / hermite interpolation / cubic spline interpolation / Chebyshev fitting

Cite this article

Download Citations
Jianmin WANG , Yabo LI , Huizhong ZHU , Tianming MA. Interpolation Method Research and Precision Analysis of GPS Satellite Position. Journal of Systems Science and Information, 2018, 6(3): 277-288 https://doi.org/10.21078/JSSI-2018-277-12

1 Introductions

GPS (global positioning system) positioning is able to obtain receiver position by calculating the distance from satellite to receiver and satellite position[1, 2], the distance that from satellite to receiver would be equal to the multiplication between propagation velocity and propagation time, thus, calculating the satellite position is a key element in GPS positioning. Given by global tracking station, GPS precise ephemeris can provide high precision and discrete satellite position[3-6], but not apply satellite position at any observation time. Therefore, how to obtain the satellite position at any observation time is of significance problem in the study of GPS positioning. The interpolation method has advantages of simple process, high computational efficiency, which can get the epoch instantaneous satellite position. Interpolation is a matter of being composed of lots of known discrete variables and the corresponding dependent variable value, besides, being established polynomial function, which can be drawn the variable value at any discrete points. There are a variety of interpolation methods in resolving satellite position, this paper mainly studies Lagrange interpolation, Newton interpolation, Hermite interpolation method, Cubic spline interpolation method, Chebyshev fitting method. Apart from, the article describes the principle of several methods, analyses the calculation precision of each method and discusses advantages and disadvantages, finally, comes to the optimal interpolation method in satellite position solution.

2 Interpolation Method

2.1 Lagrange Interpolation

In the independent variable x[a,b], f(x) has n+1 points x0,x1,,xn, and its corresponding values are y0,y1,,yn separately[7, 8]. Given an x in [a,b], we are capable of getting the corresponding estimate of x. The basic formula is as follows:
Ln(x)=i=0nyili(x).
(1)
Formula (1), Ln(x) denotes Lagrange interpolation polynomial; i denotes n+1 numbers; yi denotes the function value of point xi; li(x) stands for basic function, its formula is:
li(x)=j=0,jin(xxj)/(xixj).
(2)
Formula (2), xi, xj separately stands for different time nodes; x is to be inserted points; i and j are numbers. Given a x in [a,b], then you will get a estimate of x by means of Formula (1) and Formula (2). If there are two discrete points x0,x1 in [a,b], so Formula (1) is the linear interpolation function, if there are three discrete points x0,x1,x2, then Formula (1) is parabolic interpolation function, if there are n+1 points x0,x1,,xn, thus Formula (1) is n times polynomial function. During calculating satellite position, in a certain observation period, in order to obtain the satellite position at any epoch of the period, Formula (1) can be written as:
X(t)=i=0nXi(t)li(t),
(3)
Y(t)=i=0nYi(t)li(t),
(4)
Z(t)=i=0nZi(t)li(t).
(5)
Formula (3)(5), t expresses observation time of the interpolation node; li(t) expresses a basic function; Xi(t), Yi(t), Zi(t) respectively signify n+1 numbers corresponding satellite positions X0Y0Z0, X1Y1Z1,,XnYnZn in three directions; X(t), Y(t), Z(t) are separately satellites positions at the t moment in three directions.

2.2 Newton Interpolation

Function f(x) has n+1 discrete points x0,x1,,xn in the independent variable x[a,b] and its corresponding dependent variable values separately signify f(x0),f(x1),,f(xn), given a x in [a,b], corresponding dependent variable value is f(x)[9, 10]. The first step of this method is to find the value in the divided difference table, then x[a,b], correspondingly, Nn(x) is:
Nn(x)=f(x0)+f(x0,x1)(xx0)++f(x0,x1xn1)(xx0)(xx1)(xxn1).
(6)
In addition:
f(x0,x1xn1)=f(x,x0,x1xn1)f(x0,x1xn)xxn.
(7)
Formula (6) (7), Nn(x) indicates that n numbers interpolation polynomial; f(x0) represents function value of independent variable x0; x is arbitrary interpolation point; x0,x1xn are independent variables; f(x0,x1,,xn1) expresses divided difference of n orders.
A good characteristic of Newton interpolation method is that it never waste information in the divided difference table when we increases one discrete point for the independent variables, this algorithm are about to add the corresponding independent variables, the dependent variables and the difference to the divided difference table, which applied to Formula (6) can calculate function value of point to be inserted. Meanwhile, in the satellite position calculation, Formula (6) is derived as:
Xn(t)=f(t0)+f(t0,t1)(tt0)++f(t0,t1tn1)(tt0)(tt1)(ttn1),
(8)
Yn(t)=f(t0)+f(t0,t1)(tt0)++f(t0,t1tn1)(tt0)(tt1)(ttn1),
(9)
Zn(t)=f(t0)+f(t0,t1)(tt0)++f(t0,t1tn1)(tt0)(tx1)(ttn1).      
(10)
Formula (8)(10), t0,t1,,tn are regarded as the observation time of independent variables; t represents interpolation observation time; f(t0,t1,,tn1) represents n orders difference at observation time t; Xn(t), Yn(t), Zn(t) respectively stand for the satellite position at time t in three directions.
Though Formula (8), (9) and (10) can be calculated satellite position at any epoch.

2.3 Hermite Interpolation

ax0<x1<xnb, this method only consider that the function f(x) has uninterrupted features and f(x) has smooth features, taking a number of points in x[a,b] and the corresponding value f(x) to compose interpolation polynomial function p(x) whose function values are separately p(x0),p(x1),,p(xn) at the point x0,x1,,xn.
Duing to function f(x) and p(x) satisfy:
p(x0)=f(x0),p(x1)=f(x1)p(xn)=f(xn),p(x0)=f(x0),p(x1)=f(x1)p(xn)=f(xn).
Therefore both of them have a close degree. In the [a,b], yi=f(xi), mi=f(xi) (i has n+1 numbers), there are 2n+2 known values, thus, H2n+1(x) is as follows:
H2n+1(x)=c0+c1x+c2x2++c2n+1x2n+1.
(11)
Formula (11), x denotes an independent variable; H2n+1(x) denotes a function whose times are less than 2n+1. c0,c1,,c2n+1 denote function coefficient.
Function H2n+1(x) consists of Formula (2):
H2n+1(x)=k=0n[αk(x)yk+βk(x)mk].
(12)
Formula (12), yk denotes the satellite coordinates in three directions; mk stands for the first derivative value at each node; αk(x), βk(x) composed of the basic function lk(x) stands for 2n+1 times polynomial. We take advantage of coordinates at each node and the one order derivative value of each node in the polynomial function constructed by the method of 2.1 sections to calculate satellite position.

2.4 Cubic Spline Interpolation

The interval [a,b] is divided into n parts: a=x0<x1<xn=b. Given a point xk in [a,b] and the dependent variable value yk, if there exits a polynomial S(x) contends the following three conditions, then S(x) is called the interpolation function at the point xk.
1) Interpolation condition:
S(xk)=yk,k=0,1,,n.
(13)
Formula (13), xk represents a point in [a,b]; yk represents the known dependent variable value at the point xk; S(xk) represents the interpolation function value at the point xk.
2) Subsection condition: Within a cell interval, S(x) is a three algebraic polynomial:
S(x)=a0+a1x+a2x2+a3x3.
(14)
Formula (14), x is any point in [a,b]; S(x) stands for interpolation function; a0,a1,a2,a3 are polynomial coefficient.
3) Smooth condition: S(x)C2[a,b], that is to say, S(x) has continuous feature. Moreover, S(xk)=Mk stands for the value of S(x) at point xk, Mk stands for bending moment of S(x) at point xk. This method has three choices when adds conditions at the two ends of the interval, the article uses the first choice, in other words, the value S(x) is known at two points x0, xn, S(x0)=f0, S(xn)=fn, thus, function S(x) represents:
S(x)=(xk+1x)36hkMk+(xxk)36hkMk+1+(ykMkhk26)xk+1xhk+(ykMk+1hk26)xxkhk.
(15)
Formula (15), S(x) represents the interpolation function; xk, xk+1 stands for end points at each cell interval; Mk, Mk+1 respectively stands for bending moment of S(x) at point xk, xk+1; yk, yk+1 respectively stands for dependent variable value at pointxk, xk+1; hk denotes the length of the interval.
According to [a,b] have n+1 points and corresponding function values are f(x) and the values of function S(x) at points x0, xn, then being able to solve satellite position.

2.5 Chebyshev Fitting

This method is mainly divided into four steps to calculate the satellite position; the specific process is as follows:
1) We take the epoch interval [t0,t0+Δt] normalize into τ[1,1][11-16], normalized formula is expressed as:
τ=2(tt0)Δt1.
(16)
Formula (16), t0 stands for the first time epoch; Δt stands for the length of observation time; t stands for any observation time.
2) Taking regard X direction as an example, the satellite position at any epoch is represented as:
Xk=i=0nCxiTi(τk).
(17)
Formula (17), Xk represents the SP3 format data satellite position at interval time; Cxi denotes a polynomial unknown coefficient in the X direction time interval. Ti(τk)stands for function value of Ti(k) at τk.
The correction Vxk for satellite X direction:
Vxk=i=0nCxiTi(τk)Xk.
(18)
Here are error equations:
[Vx0Vx1Vxn]=[T0(τ0)T1(τ0)Tn(τ0)T0(τ1)T1(τ1)Tn(τ1)T0(τn)T1(τn)Tn(τn)][Cx0Cx1Cxn][X0X1Xn].
(19)
Formula (19), Vxk indicates satellite correction; the remaining variables and Formula (17) have the same meaning.
3) Calculating the coefficient matrix in the formula (19), including T0(τ)=1, T1(τ)=τ, T2(τ)=2τ21, Tn(τ)=2τTn1(τ)Tn2(τ), |τ1|, n2.
4) Formula (19) is put in order V=ACX and uses the least squares concept VTPV=min[17, 18], weight P=I, thus, C can be written:
C=(ATA)1ATX.
(20)
Formula (20), A stands for the coefficient in Formula (19); X represents a matrix composed of the satellite position at interval epoch.
After the polynomial coefficient obtained, we combine polynomial coefficient with Step 3) put into Formula (17) will reckon satellite position at arbitrary epoch. The above four steps also apply to Y and Z direction.

3 Empirical Analysis

From IGS (International GNSS service) website downloads precise ephemeris data on the 2015-4-29, this thesis adopts 0:00:00–6:30:00 time period and takes regard 30 min as time interval, satellite position at 30 min interval are known. The essay calculates total ten satellites position at 15 min by using five interpolation methods and has a comparative analysis with real satellite position at 15 min from IGS. Limited to length, the article only describes PRN01 satellite position at 15 min by making use of the interpolation method and lists the data analysis results, the rest of the satellite compared with PRN01 have the same conclusion.

3.1 Position Deviation

During 0:00:00–6:30:00 period, the interval time is 30 min, every 30 min is the independent variable node, the corresponding three dimensional satellite coordinates are the dependent variable value. The method in Subsection 2.1 calculates satellite position with increasing node numbers, then converts to E (East), N (North), U (Up) direction deviation (Table 1). Newton interpolation method choose different orders in order to calculate satellite position, then converts to E, N, U direction deviation (Table 2). Hermite interpolation method and Cubic spline interpolation's eight orders polynomial interpolation function consist of Formula (1), then we calculate one order derivative of independent variables node in the formula (12) and Subsection 2.4, meanwhile compute satellite position with different orders, then converts to E, N, U direction deviation (Table 3 and Table 4). The Chebyshev fitting method, firstly calculates the polynomial coefficients, secondly combines with step three in Subsection 2.5 to compute the satellite position, and finally gets direction deviation (Table 5).
Table 1 The deviation of satellite position in ENU direction obtained by Lagrange interpolation method
order dE/m dN/m dU/m
1 12.8325 53.9375 193.7122
2 24.3621 2.7839 6.4433
3 2.3934 1.4973 5.9420
4 1.2433 0.5435 0.7586
5 0.3336 0.1655 0.2820
6 0.0714 0.0837 0.0903
7 0.0444 0.0082 0.0111
8 0.0005 0.0112 0.0110
9 0.0054 0.0011 0.0011
10 0.0013 0.0012 0.0012
11 0.3700×103 0.5648×103 0.5485×103
12 0.3571×103 0.0822×103 0.0181×103
13 0.7252×104 0.0229×104 0.8767×104
Table 2 The deviation of the satellite position in ENU direction obtained by Newton interpolation method
order dE/m dN/m dU/m
1 12.8325 53.9375 193.7121
2 24.3621 2.7839 6.4433
3 2.3934 1.4973 5.9420
4 1.2433 0.5435 0.7586
5 0.3336 0.1655 0.2820
6 0.0714 0.0837 0.0903
7 0.0444 0.0082 0.0111
8 0.0005 0.0112 0.0110
9 0.0054 0.0011 0.0011
10 0.0013 0.0012 0.0012
11 0.3700×103 0.5648×103 0.5485×103
12 0.3571×103 0.0822×103 0.0181×103
13 0.7252×104 0.0229×104 0.8767×104
Table 3 The deviation of the satellite position in ENU direction obtained by Hermite interpolation method
order dE/m dN/m dU/m
1 0.0480 0.1079 0.4114
2 0.0013 0.0141 0.0157
3 0.0002 0.0097 0.0131
4 0.0002 0.0012 0.0293
5 0.0001 0.0927 0.1639
6 0.0007 0.7829 1.1704
7 0.0012 5.5790 8.1409
8 0.0006 36.8771 53.5668
9 0.0120 231.2324 335.4841
10 0.0001×103 1.3908×103 2.0163×103
11 0.0005×104 0.8088×104 1.1704×104
12 0.0089×104 4.5755×104 6.5934×104
13 0.0121×105 2.5316×105 3.6172×105
Table 4 The deviation of the satellite position in ENU direction obtained by cubic spline interpolation method
order dE/m dN/m dU/m
1 6.2448 9.3675 31.2143
2 6.3380 9.3841 31.1910
3 6.3245 9.3758 31.2059
4 6.3266 9.3761 31.1974
5 6.3266 9.3794 31.2047
6 6.3265 9.3751 31.1978
7 6.3265 9.3790 31.2036
8 6.3265 9.3761 31.1993
9 6.3265 9.3780 31.2021
10 6.3265 9.3768 31.2005
11 6.3265 9.3775 31.2013
12 6.3265 9.3772 31.2009
13 6.3265 9.3773 31.2011
Table 5 The deviation of the satellite position in ENU direction obtained by Chebyshev fitting method
order dE/m dN/m dU/m
1 12.8325 53.9375 193.7122
2 24.3621 2.7839 6.4433
3 2.3934 1.4973 5.9420
4 1.2433 0.5435 0.7586
5 0.3336 0.1655 0.2820
6 0.0714 0.0837 0.0903
7 0.0444 0.0082 0.0111
8 0.0005 0.0112 0.0110
9 0.0054 0.0011 0.0011
10 0.0013 0.0013 0.0013
11 0.3723×103 0.5655×103 0.5493×103
12 0.3569×103 0.0819×103 0.0185×103
13 0.7251×104 0.0225×104 0.8770×104
In Table 1, when the order is small, put another way, closing to the edge of the interpolation interval will lead to Runge phenomenon, so we must choose the appropriate interpolation order. In E direction, the accuracy of six orders reach to 0.0714 m, subsequently, the accuracy has gradually improved and interpolation effect is getting better. In N direction, it is because small interpolation order that big error and poor accuracy, the precision from eight orders arrives within 0.0112 m. In U direction, from eight orders, interpolation accuracy achieves within 0.0111 m, the error of one order is 193.7122 m and has not a good accuracy. The common point in E, N, U is that the accuracy have gradually increased from one to thirteen orders.
In Table 2, In E direction, interpolation accuracy is poor from one to five orders and has reached within 7 cm when orders are more than six. In N direction, the accuracy of above six orders reaches 8.37 cm. In U direction, interpolation precision has reached 1 cm when the order is seven, afterwards the accuracy gets higher and higher and interpolation results become better and better. The errors which are less than four orders are relatively large in E, N, U direction and can not meet the accuracy requirements. The precision of this method is equal to that of Table 1.
In Table 3, the sampling rate of Hermite interpolation is 30 min, we select nine epochs to constitute the eight orders Lagrange interpolation function in X,Y,Z three aspects, obtains one order derivative of each node in three directions. The essay makes use of epoch value and corresponding coordinates at each node and the one order derivative in X,Y,Z direction, which be able to calculate PRN01 satellite position at 15 min and translates coordinates into E, N, U direction deviation. Low order Hermite interpolation method is better than high order interpolation method, we can improve the accuracy by reducing interpolation number. In E direction, the accuracy runs up to 0.0001 m, the accuracy become worse when the number are above twenty-one. In N direction, when the numbers are less than nine, the accuracy of interpolation reaches 0.0012 m. In U direction, interpolation accuracy reaches 12 cm, the accuracy starts to decline from 1.1704 m when the interpolation number is thirteen and the accuracy decreases obviously when the number above fifteen times. The common point in E, N, U is that the better accuracy, less than nine times.
In Table 4, cubic spline interpolation follows the eight orders Lagrange interpolation function in Table 3 and obtains the one order guided value at each node, this article uses the first boundary conditions and different orders to get PRN01 satellite position at 15 min. Then we convert XYZ satellite position into ENU coordinate system and gets E, N, U deviation. The deviation obtained by this method is stable and has no Runge phenomenon, but the accuracy of E, N, U direction are 6 m, 9 m, 31 m, deviation will make small floating when order number is different, this method is not better than Table 1 and Table 2.
In Table 5, when epoch numbers have increased, deviation value in E, N, U direction calculated by Chebyshev fitting method are shrinking, from one to four orders the deviation are big and can not contend requirements, the precision has improved from five orders, In E, N, U direction precision have reached 4 cm, 8 mm, 1 cm when the order is seven and reach the seven orders precision of the Newton interpolation method.

3.2 Accuracy Analysis

Using RMS (root mean square) value can measure the pros and cons of the algorithm. The calculation formula is as follows:
RMSE=ΔE12+ΔE22++ΔEn2n,
(21)
RMSN=ΔN12+ΔN22++ΔNn2n,
(22)
RMSU=ΔU12+ΔU22++ΔUn2n.
(23)
Formula (21), (22) and (23), RMSE, RMSN, RMSU are the RMS value of E, N, U direction in respectively; ΔEn is deviation in E direction; ΔNn is deviation in N direction; ΔUn is deviation in U direction. n is the numbers of ΔE, ΔN, ΔU.
The calculation result is shown in Figure 1.
Figure 1 RMS values of different interpolation methods in ENU direction

Full size|PPT slide

In Figure 1, RMS value of cubic spline interpolation method in E, N, U direction are RMSE=6.3210, RMSN=9.3769, RMSU=31.202, so the accuracy is poor and cannot meet the requirements. Hermite interpolation method RMS value is not large, but when the order is very high, E, N, U direction deviation will become big and have poor accuracy. RMS value of the Chebyshev fitting method are RMSE=0.1147, RMSN=0.0620, RMSU=0.0988 in respectively, its accuracy is better than Hermite interpolation method. In E, N, U direction, the accuracy are consistent between Newton interpolation method and Lagrange interpolation method and are worse than the latter three methods in this table, RMSE=0.0298, RMSN=0.0300, RMSU=0.0324.

3.3 Computation Time Length

In the five interpolation methods, we can select the order which can ensure satellite position precision reach the centimeter level and obtain the operation time in respectively. The results are shown in Figure 2.
Figure 2 Computation time of different interpolation methods in ENU direction

Full size|PPT slide

In Figure 2, Newton interpolation runs faster than the first method. Hermite interpolation method runs slightly faster than the Newton interpolation in N direction, but correspondingly, RMSE=6.3210 m, RMSN=9.3769 m, RMSU=31.202 m, especially when the interpolation times between 21 and 27, E, N, U direction deviation are big and cannot meet the requirements. The Chebyshev fitting has a slightly speed. Cubic spline interpolation method cannot reach the centimeter precision, so the computing time is not considered.

4 Conclusions

During 0:00:00–6:30:00 time, the experimental results show that E, N, U direction precision reaches centimeter when the epoch number of Newton interpolation and Lagrange interpolation is more than seven. Hermite interpolation method whose times are less than eleven can reach the accuracy of eight orders Lagrange interpolation method. Although cubic spline interpolation method has no Runge phenomenon, deviations in Table 4 are not up to the results in Table 1 and Table 2. Chebyshev fitting precision has gradually became higher owing to more epoch nodes, when the order is more than seven, the precision can reach seven orders Lagrange interpolation method. On the whole, the accuracy of cubic spline interpolation method is within meters, this method is the worst. Hermite interpolations, Chebyshev fitting method, their accuracy gradually become higher. In E, N, U direction, deviation results and RMS value of Newton and Lagrange interpolation method are equivalent, its accuracy is higher than the other three methods, moreover, it is because the running time of former is faster than the latter that the former replace the latter, in a comprehensive comparison, Newton interpolation is optimal method for satellite position solution.

References

1
He F, Gao C F, Pan S G, et al. The precise point positioning of different ephemeris and clock products analysis of the results of the. Navigation Journal, 2015, 3 (4): 79- 84.
2
Zhang X H, Guo F, Li X X, et al. GPS/GLONASS combined precision single point positioning research. Journal of Wuhan University: Information Science Edition, 2010, 35 (1): 9- 12.
3
Wang J X, Wang J G, Chen J P. The satellite position and velocity of the Beidou satellite broadcast ephemeris fitting based on. Journal of Tongji University (Natural Science Edition), 2016, 44 (1): 155- 160.
4
Li Z H, Huang J S. GPS measurement and data processing. 2nd Edition Wuhan: Wuhan University Press, 2010: 73- 74.
5
Shi W P, Liu X. Generalized extended application of. Surveying and spatial information extension method in IGS precise ephemeris interpolation, 2014, 37 (11): 143- 145.
6
Sun P, Zhao C S, Wang R. A new precise ephemeris interpolation method. Global Positioning System, 2015, 40 (6): 89- 91.
7
Zhang Z J, Wang Y L, Zhang J. Two kinds of interpolation methods for precise ephemeris interpolation data processing. Beijing Surveying and Mapping, 2015, (5): 40- 43.
8
Liu Y H. Interpolation and fitting methods for Spacecraft Orbit Determination of satellite coordinates. Xi'an: Xi'an Electronic and Science University, 2014, 29- 30.
9
Tong H B, Sa H, Zhang G Z, et al. A real-time interpolation method of GNSS satellite orbit high precision. Journal of National University of Defense Technology, 2012, 34 (2): 59- 63.
10
Sun H L, Zhang Z Z. Accuracy analysis of interpolation method in the calculation of global positioning system satellite position. Journal of Navigation and Positioning, 2013, 1 (3): 89- 93.
11
Ji C D, Xu A G, Feng L. GPS processing method of precise ephemeris fitting and interpolation of the Runger phenomenon. Science of Surveying and Mapping, 2011, 36 (6): 169- 171.
12
Wang X, Gao J X, Wang J, et al. Using Chebyshev polynomial fitting sliding coordinates and satellite precise clock. Bulletin of Surveying and Mapping, 2015, (5): 6- 8+16.
13
Wang W, Chen M J, Yan J Q, et al. Three kinds of Beidou satellite precise ephemeris interpolation methods to compare and analyze the. Global Positioning System, 2016, 41 (2): 60- 65.
14
Peng X Q, Gao J X. Broadcast ephemeris interpolation sliding Chebyshev analysis method of. Technology Based on Coal, 2015, 34 (6): 104- 106.
15
Chen P, Chen Z Y, et al. A new method of GPS precise ephemeris interpolation. Geodesy and Geodynamics, 2009, 29 (2): 111- 114.
16
Wang J M, Huang J P, Zhu H Z, et al. Ionospheric total electron number forecast method research. Science of Surveying and Mapping, 2016, 41 (12): 47- 53.
17
Wang J M, Ma T M, Zhu H Z. Fast resolution of double frequency BDS integer ambiguity realized by improved LAMBDA algorithm. Systems Engineering-Theory & Practice, 2017, 37 (3): 768- 772.
18
Wang J M, Ma T M, Zhu H Z. Real time rapid resolution of integer ambiguity for BDS/GPS combined system. Journal of China University of Mining & Technology, 2017, 46 (3): 672- 678.

Funding

the National Natural Science Foundation of China(41474020)
PDF(182 KB)

307

Accesses

0

Citation

Detail

Sections
Recommended

/