Process RTK Data Based On Kalman Filtering Algorithm

International Journal of Research in Engineering and Science (IJRES)
ISSN (Online): 2320-9364, ISSN (Print): 2320-9356
www.ijres.org Volume 3 Issue 5 ǁ May 2015 ǁ PP.39-45
www.ijres.org 39 | Page
Process RTK Data Based On Kalman Filtering Algorithm
Yi Guan-min 1
and Zhou Zhi-feng2
1
College of Mechanical Engineering, Shanghai University of Engineering Science, Shanghai 201620, China
2
College of Mechanical Engineering, Shanghai University of Engineering Science, Shanghai 201620, China
ABSTRACT: With the development of satellite positioning technology, there is a strong need for high accuracy
position information. Currently the most widely used high-precision positioning technology is RTK(Real-Time
Kinematic).RTK technology is the key to using carrier phase measurements. It takes advantage of the base
stations and monitor stations observed error of spatial correlation, except monitor stations observed by means of
differential most of the errors in the data, in order to achieve high accuracy positioning.[3]
Based on Kalman
filtering algorithm to handle the noise of RTK data and selecting appropriate models to further improve the
accuracy of the data. This paper will explore the use of Kalman filtering method of RTK data processing, which
reduces random noise interference, thus improving the accuracy of GNSS deformation monitoring data.[1]
KEY WORDS: Kalman filtering; RTK; Carrier phase measurement; High-accuracy
I. INTRODUCTION
Kalman filtering using the time-domain state-space method, applicable to Multivariable systems with
time-varying systems and non-stationary random process, has a minimum of the unbiased variance.It is
currently the most widely used method of dynamic data processing because of its recursive feature easy to
implement.[1]
Through the establishment of State equations and equations to describe the system of measurement
of the dynamic process of change according to the filter gain matrix, quantitative identification and extraction of
information from measurements, revision status parameter, suitable for real time data processing.
High precision GNSS measurements must be in the carrier phase observations. RTK positioning technology
is real time kinematic positioning technology based on carrier phase observations. [2]
It can provide a logging site
in real time three dimensional positioning result in the specified coordinate system and achieve centimeter-level
accuracy. In RTK operation mode, the base station will transmit its observation and monitoring station
coordinates information to the monitor station.[3]
Monitor station not only via the data chain receives data from
the base station but also gathered a GNSS observation data. It is real-time system which processes differential
measurements. Centimeter-level positioning is also given. Fixed the ambiguity of whole cycles, you can conduct
real-time processing of each epoch. Just keep track four or more Satellites and necessary geometry, monitor
station may at any time give a centimeter-level positioning results.
II. KALMAN FILTERING PRINCIPLE AND ALGORITHM
Assume that )(
~
kS is the estimation of signal S(k). )|1(
~
kkS  is the S(k+1)’s forecast estimate in the time
of k.Then the model of AR(N)[1][4][5]
is:

)()1()( kWkSkS  (1)
Observations X(k) are noise and signal superposed,
)()()( kNkCSkX  (2)
The criterion for estimating:
min)]([ kTr  ;
min)]1([ kTr  .
Such as
]))|1(
~
)1())(|1(
~
)1([()1( T
kkSkSkkSkSEk 
Filter gain:
1
])1|()([)1|()( 
 T
n
T
CkkCkVCkkkB 
(3)
Filter estimate:
)]1(
~
)()[()1(
~
)(
~
 kSCkXkBkSkS (4)
Update status: )(
~
)|1(
~
kSkkS  (5)
Estimated standard deviation:
Filtering:
)1|()()1|()(  kkCkBkkk  (6)
Forecast:
)()()|1( kVkkk T
 
(7)
In practical applications, a lot of times we get random signal measurements, without the signal model. However,
we can use the observations X(k) to estimate the model. This algorithm uses the Burg method of the model of
AR(P) .[4][5]
Suppose
)1|( pnef
、
)1|( pneb 、
)1( p
 known：
)1|( pnef : (P-1)-order Predictor of forward prediction error.
)1|( pneb : (P-1)-order Predictor of backward prediction error.
)1( p
 : Predictor of the variance of the prediction error.
Use time values instead of reflection coefficient which its total statistical mean value estimation model of order
p:










 1
1
2
1
1
2
1
1)(
)1|1()1|(
)1|1()1|(2
N
pn
b
N
pn
f
N
pn
bf
p
pnepne
pnepne
P (9)
By recursive formula which belong to auto-correlation algorithm for estimation of that p-model’s coefficient and
white noise’s power :
)()( pp
pa  (10)

)()1(1)( pp
kp
p
k
p
k eaaa 



k=1～p (11)
])(1[ 2)()1()( ppk
  
(12)
Recursive high-order model prediction error sequences:
)1|1()1|()|( )(
 pnepnepne b
p
ff  1 Npn ～ (13)
)1|()1|1()|( )(
 pnerpnepne f
p
bb
(14)
Initialization is used to calculate:
)()0|()0|( nxnene bf 
n=0～N-1
)(
1 1
0
2
0
)0(
nx
N
R
N
n




In addition, at the start of the mean-square deviation from the actual situation through experiment.When
)(m
is less than the requirement, stop counting.
III. APPLICATION OF KALMAN FILTER IN GNSS DEFORMATION MONITORING CASE STUDY
In this case, stochastic signal model：
)()2()1()( 21 kkSakSakS  (16)
Variance of white noise is
2
w . Vectorization of the above formula ,we can get
)()1()( kWkSkS  (17)
Assume that








)1(
)(
)(
ks
ks
ks
；







01
21 aa
；







0
)(
)(
k
kW

Observations x(k) are the sum of the signal s(k) and n(k) noise of variance measurement
2
n :
)()()( knkskx  (18)
Assume that
)]([)( kxkX  ；
]01[C ；
)]([)( knkN  .
We can get
)()()( kNkCSkX 
Covariance matrix of random white noise：








00
0
)]()([)(
2
wT
w kWkWEkV

； ][)]()([)( 2
n
T
n kNkNEkV 
By equation (3) and (4) we know that
Filter gain matrix:







)(
)(
)(
2
1
kb
kb
kB
Filter estimate matrix:
 









































)]1|(
~
)()[()2|1(
~
)]1|(
~
)()[()1|(
~
)2|1(
~
)1|(
~
01)(
)(
)(
)2|1(
~
)1|(
~
)]1(
~
)()[()1(
~
)(
~
2
1
2
1
kkSkxkbkkS
kkSkxkbkkS
kkS
kkS
kx
kb
kb
kkS
kkS
kSCkXkBkSkS
So,
)]1|(
~
)()[()1|(
~
)(
~
1  kkSkxkbkKSkS (20)
)]1|(
~
)()[()2|1(
~
)1(
~
2  kkSkxkbkKSkS (21)
We can get the equation (22) by the next prediction matrix )|1(
~
kkS 





 





















)(
~
)1(
~
)(
~
)1(
~
)(
~
01)1|(
~
)|1(
~
2121
kS
kSakSa
kS
kSaa
kkS
kkS
(22)
)(
~
)|1(
~
KSkkS  (23)
So,
)1(
~
)(
~
)|1(
~
21  kSaKSakkS (22)
Next observation value should satisfy the following equation
)|1()1()|1(
~
kkkxkkS  
(23)
The expression of )|1( kk 
)|1()|1( kkkk  

IV. EXPERIMENT AND CONCLUSION
We are using M300C&M300C-U receivers of Shanghai compass satellite navigation company in this
experiment. These receivers are multi-satellite measurement receivers which based on Beidou satellite
navigation system using RTK unique core technologies and highly reliable carrier tracking algorithm in order to
provide a comparison of quality of results. The experimental data is collected on the spot in our University using
single reference station network. The base station settings in the roof of our training, monitor Station fixed on a
slope at the campus. Diagram of base station is given in Figure 4.1. It consist of four parts such as Receiver、
Radio、Receiver antenna and Radio antenna. Monitor station consist of Receiver which include radio、Receiver
antenna and radio antenna.
Figure 4.1 The composition of Base Station
Star charts and PVT views are shown in Figure 4.2. Through this diagram we can see that Satellite tracking and
the number of visible satellites and so on. These information shown that the quality of GNSS signal is pretty
good.
Figure 4.2 Star charts and PVT views
RTK data collected by this experiment and the enactment of this algorithm for filtering the data drawing are
that shown in Figure 4.3、4.4 and 4.5.[6]

Figure 4.3 The result of processing RTK data using Kalman Filter algorithm in WGS-84 (direction X)
Figure 4.4 The result of processing RTK data using Kalman Filter algorithm in WGS-84 (direction Y)
Figure 4.5 The result of processing RTK data using Kalman Filter algorithm in WGS-84 (direction Z)
The accuracy of the receiver of M300C is 2.5cm. From the above diagrams have shown that accuracy of
RTK date through Kalman filtering have Significantly improved, especially in direction of x and y.

Millimeter-scale positioning accuracy is achieved. In the direction of z, however, the date accuracy is about
1cm.Through the above described we found that we can use RTK in many cases, such as disaster monitoring.
REFERENCES
[1] Qin Yongyuan, Zhang Hong, Wang Shuhua, Kalman filtering and navigation theory,2nd Edition[M]. Publishing house of
Northwestern Polytechnical University, 2012, pp1-3, pp33-57.
[2] Ren Dongyuan, Zhang Lin,(2014) Error analysis of RTK surveying[J], Private science and technology journal, No.11, pp68-69.
[3] Ji Jing, Zheng Zhi-Jiang, Chen Fu-Chao,(2014)Forecast and inspection of land subsidence disaster in Binhai new area of
Tianjin[J], Shanghai Land & Resources,Vol.35, N0.4,pp137-141
[4] Xia Zhihao, Zhao Chang-sheng, Ma Xiao-jun,(2015)GPS Single Point Positioning Based on Adaptive Kalman Filter of the
Colored Noise[J], Bulletin of surveying and mapping, No.3,pp66-68.
[5] William W.S. Wei, TIME SERIES ANALYSIS Univariate and Multivariate Methods,2nd Editor[M]. Publishing house of
Addison Wesley,pp30-50,pp435-456.
[6] Edward B. Magrab, Balakumar, Balachandran, Keith E.Herold, Shapour Azarm, James H.Duncan, Gregory C.Walsh, An
Engineer’s Guide to MATLAB: with application from mechanical,Aerospace,Electrical,and Civil Engineering,Second Edition
[M]. Publishing house of Electronics Industry,pp159-190.
ACKNOWLEDGEMENT
The author would like to thank Professor Zhou Zhifeng who helped and supported with this work. This wor
k was supported by Shanghai University of Engineering Science postgraduate innovation project(No.14KY013).
Authors
Corresponding author: Yi Guanmin (1989-), male, master of engineering, research
measurement & control. E-mail:yiguanmin@126.com

Process RTK Data Based On Kalman Filtering Algorithm

More Related Content

What's hot (20)

Similar to Process RTK Data Based On Kalman Filtering Algorithm (20)

More from IJRES Journal (20)

Recently uploaded (20)

Process RTK Data Based On Kalman Filtering Algorithm