1
Dr. Guoqing Zhou
9. Data Processing (1)
CET 318
Book: p. 203-276
1. Data Pre-Processing
1.1 Data Handling
1. Downloading
2. Data Management
3. Data Exchange
1. RINEX format consists of:
(1) The observation data file containing the range data;
(2) The meteorological data file; and
(3) The navigation message file.
2. RINEX Data Format
1. RINEX observation data file
Header section
Data section
2. RINEX meteorological data file
Header section
Data section
3. RINEX navigation message file
Header section
Data section
1.2 Cycle Slip Detection and Repair
1. Definition of Cycle Slips
During tracking, at a given epoch, the observed accumulated
phase ??
∑∑
=
=
=
=
++=
titi
N
1i1i
2π? ??
The initial integer number N of cycles between the satellite
and the receiver is unknown. This phase ambiguity N
remains constant as long as no loss of the signal lock occurs.
In this event, the integer counter is re-initialized which
causes a jump in the instantaneous accumulated phase by an
integer number of cycles. This jump is called cycle slip.
2. Causes of Cycle Slips
1. Obstructions of the satellite signal due to trees, buildings,
bridges, mountains, etc. This source is the most frequent
one.
2. Low SNR due to bad ionospheric conditions, multipath,
high receiver dynamics, or low satellite elevation.
3. Failure in the receiver
software, which leads to
incorrect signal processing.
4. Malfunctioning satellite
oscillators, but these cases are
rare.
3. Testing Quantities
The formulation of testing quantities is based on measured
carrier phases and code ranges.
Phase and integrated
Doppler combination
Single frequency phase and
Doppler
Phase and code range
combination
Single frequency phase and code
range
Phase combinationsDual frequency phases (L1 and L2)
Raw phaseSingle frequency phase (L1 or L2)
Testing quantityRequired data
Single receiver tests enable in situ cycle slip detection and
repair by the internal software of the receiver.
For a single site, the testing quantities are
2
4. Detection and Repair
Each of the described testing quantities allows the location of
cycle slips by checking the difference of two consecutive epoch
values. This also yields an approximate size of the cycle slip.
One of the methods for cycle slip detection is the scheme of
differences.
1. Detection of Cycle Slip
Interpolation Techniques: A method is to fit a curve through
the testing quantities before and after the cycle slip.
Prediction Method: At a certain epoch, the function value (i.e.,
one of the testing quantities) for the next epoch is predicted
based on the information obtained from preceding function
values.
2. Determination of Size of Cycle Slip
Kalman filtering
Linear regression
Least square method
3. Repair of Cycle Slip
Cycle slip repair using the ionospheric residual are
1. Based on the measurement noise assumption, the
separation of the cycle slips is unambiguosuly
possible for up to ±4 cycles.
2. A smaller measurement noise increase the
separability.
3. For a larger cycle slips, another method should be
used in order to avoid wrong choices in ambiguous
situation.
2. Ambiguity Resolution
2.1 General Aspects
Iono
λ
1
-Nδfρ
λ
1
?+?+=Φ
The ambiguity inherent with phase measurements depends
upon both receiver and satellite. The model for phase is
If we consider the ambiguity, N as an integer value, the
ambiguity is said to be resolved or fixed.
Ambiguity fixing strengthens the baseline solution, but
sometimes solutions with fixed ambiguities (i.e., integer
values) and float ambiguities (i.e., real values) may agree
within a few millimeters
1. Ambiguity Measurement The use of double-differences instead of single-differences
for carrier phase processing can eliminate the clock terms
and the isolation of the ambiguities is possible.
Why?
2. Double-Differences
For high accuracy of the carrier phase observable, the
ambiguities must be resolved to their correct integer value
since one cycle on the L1 carrier may translate, in the
maximum, to a 19 cm position error. It should be stressed
here that integer ambiguity resolution may not always
be possible.
3. Effect of Ionosphere, Troposphere and Other
Minor
3
Satellite Geometry: The number of satellites tracked at
any instant translates into a better dilution of precision
value. Thus, if a receiver tracks seven or eight satellites, it
is preferable since redundant satellites aid in the efficiency
and reliability of ambiguity resolution.
Length of Measurement Time: The information content
of the carrier phase is a function of time which is directly
correlated to the movement of the satellite.
4. Satellite Geometry and Measurement Time
The time is a critical component of ambiguity resolution
even under good geometric conditions.
Example: every 15 seconds for one hour; every second for
four minutes, a total of 240 measurements per satellite.
Since multipath is station dependent, it may be significant for
short baselines. As in the case of atmospheric and orbital
errors for long baselines, multipath has the effect of both
contaminating the station coordinates and ambiguities
6. Major Steps of Ambiguity Resolution
Multipath is also a critical factor for ambiguity resolution.
5. Multipath and Length of Baseline
The Main First Step is the generation of potential integer
ambiguity combinations that should be considered by the
algorithm.
A combination is comprised of an integer ambiguity for,
e.g., each of the double-difference satellite pairs.
Search Space is the volume of uncertainty which
surrounds the approximate coordinates of the unknown
antenna location.
Static positioning: it can be realized from the so-
called float ambiguity solution,
Kinematic positioning: it is realized from a code
range solution.
Size of the Search Space will affect the efficiency, i.e.,
computational speed.
A larger search space gives a higher number of potential
integer ambiguity combinations to assess, which in turn
increases the computational burden. This is typically
important for kinematic applications. It is necessary to
balance computational load with a conservative search space
size.
The Second Major Step is the identification of the
correct integer ambiguity combination.
The Third Major Step is the a validation (verification)
of the ambiguities.
2.2 Basic Approaches
1. With Single Frequency Phase Data
For only one frequency (L1 or L2) available, the most direct
approach is.
1. Establishing model by Eq. (9.14),
2. Linearizing the model,
3. Solving the model, a number of unknowns (e.g., point
coordinates, clock parameters, etc.) is estimated along
with N in a common adjustment.
When using double-differences over short baselines, this approach is
usually successful. The critical factor is the ionospheric refraction
which must be modeled and which may prevent a correct resolution
of all ambiguities.
1. The unmodeled errors affect all estimated parameters.
Therefore, the integer nature of the ambiguities is lost and
they are estimated as real values.
2. To fix ambiguities to integer values, a sequential
adjustment could be performed.
3. After an initial adjustment, the ambiguity with a computed
value closest to an integer and with minimum standard
error is considered to be determined most reliably. This
bias is then fixed, and the adjustment is repeated (with one
less unknown) to fix another ambiguity and so on.
Characteristics of Single-Frequency:
4
2. With Dual Frequency Phase Data
Why Dual Frequency: because of the various possible linear
combinations.
f
L1-L2
= 347.82 MHz, λ
L1-L2
= 86.2 cm
λorg = 19~24.4 cm
Wide Lane and Narrow Lane Techniques:
21 LLw
Φ?Φ=ΦWide Lane Signal:
Significant
increase compared
to the original
wavelengths
The increased wide lane wavelength provides an increased
ambiguity spacing.
This is the key to easier resolution of the integer ambiguities.
2121 LLLL
Φ?Φ=Φ
?
2121 LLLL
fff ?=
?
2121 LLLL
NNN ?=
?
The adjustment based on the wide lane model gives wide lane
ambiguities N
L1-L2
which are more easily resoled than the base
carrier ambiguities.
)1()(
2
1
21121
1
212111
L
L
LLLLL
L
LLLLLL
f
f
f
b
f
b
f
f
NN ??+?Φ?Φ=
??
??
To compute the ambiguities for the measured phases, we can get
What is b? please see p. 105, Eq.6.73
21
21
21
1
212111
)(
LL
LL
LL
L
LLLLLL
ff
ff
b
f
f
NN
+
+?Φ?Φ=
?
??
After
ionospheric
Geometry-free linear
phase combination
The disadvantage of this combination is
The corresponding ambiguity is no longer an
integer.
The ionosphere is a problem or the ionospheric
influence is eliminated which destroys the integer
nature of the ambiguities.
The use of other linear combinations ranging from
narrow lane with a 10.7 cm wavelength to extra wide
with a 172.4 cm wavelength.
Characteristics of Dual-Frequency:
3. By Combining Dual Frequency Carrier Phase
and Code Data
The most unreliable factor of the wide lane technique is
the influence of the ionosphere which increases with
baseline length. This drawback can be eliminated by a
combination of phase and code data.
1
1
11
N
b
a
L
L
LL
f
f +?=Φ
1
11
b
a
L
LL
f
fR +=
carrier phases
code ranges
2
2
22
N
b
a
L
L
LL
f
f +?=Φ
2
22
b
a
L
LL
f
fR +=
What is a,
b? please
see p.
105,
Eq.6.74
4 equations contain 4 unknowns, geometry
term, a and ionosphere term b and
))((
21
21
21
2121 LL
LL
LL
LLLL
RR
ff
ff
N +
+
?
?Φ=
??
This rather elegant equation allows for the determination
of the wide lane ambiguity N
L1-L2
for each epoch and each
site. It is independent of the baseline length and of the
ionospheric effects.
By a series derivation, we finally get
Note that even if all systematic effects cancel out, the
multipath effect remains and affects phase and code
differently.
4. By Combining Triple Frequency Carrier
Phase and Code Data
This technique for ambiguity resolution based on three
carriers is denoted as Three-Carries Ambiguity
Resolution-TCAR.
1
1
11
N
b
a
L
L
LL
f
f +?=Φ
1
11
b
a
L
LL
f
fR +=
carrier phases
code ranges
2
2
22
N
b
a
L
L
LL
f
f +?=Φ
2
22
b
a
L
LL
f
fR +=
What is a,
b? please
see p. 105,
Eq.6.74
5
5
55
N
b
a
L
L
LL
f
f +?=Φ
5
55
b
a
L
LL
f
fR +=
Similarly
6 equations contain 5
unknowns
5
3. Search Techniques
1. A Standard Approach
When processing the data based on double-differences by least
squares adjustment, the ambiguities are estimated as real or
floating point numbers.
The first double-difference solution is called the float
ambiguity solution. The output is the best estimate of the
station coordinates as well as double-difference ambiguities.
Shore baseline (e.g., 5km), and long observation span (1 hr),
the float ambiguities close to integers.
Ambiguity resolution in this case is merely used to refine the
achievable positioning accuracy.
For example:
2. Ambiguity Resolution On-the-Fly
(AROF, OTF, OTR)
Code ranges are used to define the search space for the
kinematic case. A relative code range position is used as the
best estimate of antenna location, and the associated standard
deviations are used to define the size of the search space (a
cube, a cylinder, or an ellipsoid).
To reduce the number of integer ambiguity combinations, the
code solution should be as accurate as possible which means
that receiver selection becomes important.
Low noise,
Narrow correlator-type code ranges
They have a resolution at 10 cm range and improved multipath
reduction compared with standard C/A-code receivers.
The OTF techniques have common features like,
e.g., the determination of an initial solution. A
summary of the main features is given in Table 9.5 p.
228.
Please see the Table 9.5, p. 228.
3. Ambiguity Function Method
)(δN)(ρ
1
AB
jj
AB
tft
ABAB
j
?+=Φ
λ
)(δN)(ρ
1
AB
jj
AB
tft
ABAB
j
?=?Φ
λ
)(2ππf)(πN2{
)}(ρ
λ
2π
)(πΦ2{
AB
j
AB tti
tti
j
AB
j
AB
ee
?
?
=
)}(πfδ2
πN2
)}(ρ
λ
2π
)(πΦ2{ j
AB
j
AB
ti
i
tti
eee
j
AB
j
AB ?
?
=
αsinαcos
iα
ie +=
)(πfδ2
)}(ρ
λ
2π
)(πΦ2{
AB
j
AB
ti
tti
ee
j
AB ?
?
=
For one epoch and one satellite
The basic principle by Counselman and Gourevitch is
4. Least Squares Ambiguity Search
Technique
Basic principles: is the separation of the satellites into a
primary and a secondary group.
1. The primary group consists of four satellites, which should
have a good PDOP, the possible ambiguity sets are
determined.
2. The remaining secondary satellites are used to eliminate
candidates of the possible ambiguity sets.
Approximation: This technique requires an approximation for the position
(due to the linearization of the observation equation) which may be obtained
from a code range solution.
Search Area: The search area may be established by surrounding the
approximate position by a 3δ region.
6
5. Fast Ambiguity Resolution Approach
(FARA)
The Main Characteristics of FARA
(1) to use statistical information from the initial adjustment to select
the search range,
(2) to use information of the variance-covariance matrix to reject
ambiguity sets that are not acceptable from the statistical point of
view, and
(3) to apply statistical hypothesis testing to select the correct set of
integer ambiguities.
The FARA Algorithm
1. computing the float carrier phase solution,
2. choosing ambiguity sets to be tested,
3. computing a fixed solution for each ambiguity set, and
4. statistically testing the fixed solution with the smallest
variance.
In The First Step: real values for double-difference
ambiguities are estimated based on carrier phase
measurements and calculated by an adjustment procedure
which also computes the
Cofactor matrix of the unknown parameters
Posteriori variance of unit weight (a posteriori variance
factor)
Variance-covariance matrix of the unknown parameters
Standard deviations of the ambiguities
In The Second Step: the criteria for selecting possible
ambiguity ranges (set) based on confidence intervals of the real
values of the ambiguities.
(1) First Criterion: The quality of the initial solution of the
first step affects the possible ambiguity ranges. In more
detail, the search range for this ambiguity is kδ
N
(k from
Student's t-distribution).
(2) A Second Criterion: the use of the correlation of the
ambiguities.
2
NNN
2
NN
jjiiij
δδ2δδ +?=
jiij
NNN ?=
This criterion significantly reduces the number of possible
integer sets. An even more impressive reduction is achieved if
dual frequency phase measurements are available.
In The Third Step: least squares adjustments with fixed
ambiguities are performed for each statistically
accepted ambiguity set yielding adjusted baseline
components and a posteriori variance factors.
In The Fourth Step:
1. The solution with the smallest a posteriori
variance is further investigated.
2. The baseline components of this solution are
compared with the float solution.
3. If the solution is compatible, it is accepted.
4. The compatibility may be checked by a X
2
-
distribution which tests the compatibility of the a
posteriori variance with the a priori variance.
5. Fast Ambiguity Search Filter (FASF)
FASF is comprised of basically three components:
(1)A Kalman filter is applied to predict a state vector
which is treated as observable,
(2)The search of the ambiguities is performed at every
epoch until they are fixed, and
(3)The search ranges for the ambiguities are computed
recursively and are related to each other.
6. Least Squares Ambiguity Decorrelation
Adjustment
Teunissen proposed the idea and further developed the least
squares ambiguity decorrelation adjustment (LAMBDA).
Understanding the principle of this method must have a
strong background in linear algebra.
Interested students can read through from p. 237-244
?
?
?
?
?
?
?
?
1221
1211
QQ
QQ
?
?
?
?
?
?
?
?
12
11
0
0
W
W
decorrelated
7
7. Ambiguity Determination with Special
Constraints
Several multiple receiver methods for kinematic applications exist.
1. One common procedure of this technique is to place two or
more receivers at fixed locations (usually short distances apart)
of the moving object.
2. Since the locations of the antennas are fixed, constraints (e.g.,
the distance between two antennas) may be formulated which
can be used to increase the efficiency of the ambiguity
resolution.
In principle, the gain by using constraints results in a reduction of
the potential ambiguity sets.
Background Knowledge:
4. Ambiguity Validation
After the determination of the integer ambiguities, it is
of interest to validate the quality of the obtained
quantities. There, the uncertainty of the estimated
integer ambiguities is to be determined.
How do we validate the ambiguity?
Ambiguity success rate
For more information, please read p.247-248.
4. Ambiguity Validation
Summary
What have we learnt?
Which parts are important?
Assignment 9
1. What is the data format of RINEX? Give the detailed
description (15 points)
2. What is cycle slip? What causes the cycle slip? How do
we test the quantities of the cycle slip? How do we repair
the cycle slip? (20 points)
3. List the basic approaches of ambiguity resolution (15
points).
4. Why do we use dual-frequency phase data to solve the
ambiguities (15 points).
5. How many approaches does the determination of research
space include? What are their advantages and
disadvantages (20 points)?
6. What is the ambiguity determination of constraints (15
points)?