COLE NATIONALE DE L'AVIATION CIVILE

Course: Master of Science in GNSS

DIFFERENTIAL GLOBAL NAVIGATON

SATELLITE SYSTEMS
- Report -

Student: CIUBAN Sebastian

Year: 2015/2016

TOULOUSE

TUTORIAL 2
Session A: Atmospheric effects
A.1 Differential positioning of EBRE CREU receivers (Long baseline: 288 km)
Questions (pg.18) :

What is the expected accuracy of the computed coordinates (in absolute

positioning)?

Why are the patterns seen for the receivers EBRE and CREU are so similar?

Answers:

The expected accuracy of the computed coordinates is around meter level, which
is specific for Standard Point Positioning (SPP) method. This is due to the
contribution of the important error sources like: code noise, satellite clock offset,
satellite ephemeris error (broadcasted ephemeris are used instead of precise),
atmospheric effects.

The patterns between the EBRE and CREU receivers have a similar behavior
because of the common error sources that affect their observations. Although they
are 288 km apart from each other, the common error sources cited above have
similar contributions.

Question (pg.20) :

Is the combination equivalent to use $14? Why ?

Answer:

Yes is equivalent with using $14 column from the generated file
D_EBRE_CREU.obs. This is because the combination $5 - $9 - $10 - $11
corresponds to P1 - - Tr - Iono, which gives a result equivalent with the
prefit vector from column $14.

Questions (pg.24) :

Looking at these results, justify intuitively why the errors are reduced in
differential positioning.
1

Why is the error lower in static than in kinematic mode?

Answer:

The errors are reduced in differential positioning because the common error
sources between the two stations are mitigated or they cancel.

The errors are lower in static mode because the Kalman Filter takes into account
that the receivers are not moving, so the coordinates are not changing over time.
The epoch n will be the same as n+1 from the coordinates point of view.

Question (pg.26) :

Justify the pattern seen in the nominal (Klobuchar) ionospheric corrections.

Answer:

The ionosphere is more active during the day and we can see an increase in the
differential corrections during the middle of the day. The reason we have the
sine like lines is due to the satellites which are in low elevations and high
elevations with respect to the receivers. Ionospheric delay is elevation dependent.
Low elevations will cause higher delays.

Question (pg.28) :

Taking into account the previous plot of the differential ionospheric corrections
(slide #26), discuss the effect seen on the position domain.

Answer:

In comparison with the previous results, not taking into account the ionospheric
delay we can observe a slight accentuated scatter in the middle of the day (40000
60000 GPS seconds) on the Up direction. But the effect is under meter level.

Question (pg.29) :

Discuss the pattern seen in the figure. Why is the effect of the differential
ionospheric error higher in these static positioning results?

Answer:

The effect of the differential ionospheric error is higher in the static plot results
because the error is being absorbed in the coordinates. The Kalman Filter has
memory and it propagates the error in the position.

Question (pg.31) :

Justify the pattern seen in the nominal tropospheric corrections.

Answer:

The troposphere varies locally due to the different influences (rain, clouds,
humidity, etc.) and that is why the differential tropospheric effect is 1-2 order of
magnitude bigger than ionosphere. Like the ionosphere, the tropospheric delay is
elevation dependent. This is the reason of the curved line behavior of the
corrections. High elevation satellites will have reduced delays on their transmitted
signals.

Questions (pg.33) :

Which error source has the higher impact on the position domain, the ionosphere
or the troposphere?

Which is easier to model?

Answers:

The troposphere has the highest impact on the position domain.

The troposphere is easier to model. The delay can be splitted in two components:
hydrostatic and wet. Hydrostatic component or dry component contains 90% of
the error contribution and the wet component has 10% but is more difficult to
model due to the local atmospheric parameters dependency.

Question (pg.34) :

Compare the results with those of the previous case, when not considering the
ionospheric error.

Answer:

Neglecting the tropospheric delay will affect all 3 directions while the ionosphere
affects mainly the vertical component.

A.2 Differential positioning of GARR MATA receivers (Short baseline: 51 km)

Questions (pg.40) :

Compare the results with the previous case with a baseline of 280 km.

Are the patterns even more similar now? Why ?

Answers:

The patterns are more similar between GARR MATA receivers than EBRE
CREU because the baseline is 51 km now.

The shorter the baseline the similarity between errors increases. The errors are
proportional to the baseline.

Questions (pg.44) :

Compare the results with the previous case with a baseline of 280 km.

Are we reaching a similar level of accuracy? Why ?

Answers:

An improvement is present in the position domain.

Yes, even better and that is because the errors are proportional to the baseline, in
this case the baseline being shorter, and they cancel even more. It can be observed
that for Kalman kinematic solution the errors are less scattered and for static
mode, the errors are smaller in special for vertical component.

Question (pg.46) :

Justify the pattern seen in the nominal (Klobuchar) ionospheric corrections.

Compare the plot with that of the 280 km baseline. Do the results perform as
expected?

Answer:

The receivers being closer to each other, their observations are affected by similar
ionospheric delays. Thus, the delay is well mitigated in differential positioning.
4

Nevertheless, the ionospheric influence can be observed that has an accentuated

behavior in the middle of the day

Question (pg.48) :

Do the results confirm the expected effect of neglecting the ionospheric

corrections for the baseline (using code measurements) at the user level?

Answer:

With errors around tens of centimeter level we cant identify the effect of
differential ionospheric delay at the user level if the solution is obtained in
kinematic node. Also, the code noise is too high for any identification.

Question (pg.49) :

Do the results confirm the expected effect of neglecting the ionospheric

corrections for the baseline (using code measurements) at the user level?

Answer:

As in the solution obtained in Kinematic mode, with few tens of centimeter level
error is hard to identify the effect of the differential ionospheric effect at the user
level in static mode. Small differences can be identified on Up direction between
0 10000 GPS seconds time window.

Question (pg.51) :

Justify the pattern seen in the nominal tropospheric corrections. Compare the plot
with that of the 280 km baseline. Do the results perform as expected?

Answer:

Although the baseline is shorter and the errors are smaller than previous case
when the baseline was larger, the tropospheric delay still cause around 5 meter
errors. The importance of this effect can be concluded.

Question (pg.53) :

Do the results confirm the expected effect of neglecting the tropospheric

corrections for the baseline (using code measurements) at the user level?
5

Answer:

Neglecting the troposphere will produce a negative impact at the user level as it
can be seen in the solution obtained in kinematic mode. The influence is higher on
the vertical component in this case, the errors being more scattered.

Questions (pg.54) :

Discuss the error found at the user level for the 50 km of baseline.

Is the error level similar to the error seen with the 280 km of baseline ?

Should the tropospheric corrections still be applied for these baselines?

Answers:

In the static solution we can see that the troposphere produces 10-20 cm errors in
the North and East direction while on the Up direction is about 70 cm.

The errors are not similar with the case where the baseline is 288 km. In the case
of short baseline, the tropospheric behavior has a higher probability to have the
same effect on observations and thats why mitigating it is better for 58 km
baseline.

The tropospheric model for its delays should be applied all the time.

Session B: Orbit error effects

B.1 Differential positioning of EBRE CREU receivers (Long baseline: 288 km)
Questions (pg.65) :

Is the impact on user positioning error similar for both receivers ?

From these plots, what is the expected level of error in differential positioning ?

Answers:

Yes, the impact on user positioning error is similar for both receivers. Everything
that happens in the observed satellite affects the users. As the satellite PRN17
ephemeris were corrupted between the time window 65000 8000 GPS seconds
we expect the degradation in positioning to happen in the same time window.

In this plots, we have around 200 meters error on the vertical component and
around 150 meters in horizontal one. In differential positioning it is expected to
have these errors reduced at meter level.

Question (pg.68) :

Assuming an orbit error of 2000 m, compute a threshold for the differential

error?

Answer:

In differential mode the errors in positioning reduced from 100-200 m to 10-15 m.

conservative bound

= 2000 m orbit error

b = 288000 m baseline length
= 20000000 m approximate Sat Rx distance
Result: < 28.8 m
B.2 Differential positioning of GARR MATA receivers (Short baseline: 51 km)
Questions (pg.73) :

Is the impact on user positioning error similar for both receivers ?

From these plots, what is the expected level of error in differential positioning ?

Answers:

As both receivers observed the same satellite which had its ephemeris corrupted
in the time window 65000 80000 GPS seconds, the user positioning errors are
similar.

The baseline being shorter, after we obtain the position using differential method
we expect the errors to be mitigated.

Question (pg.76) :

Assuming an orbit error of 2000 m, compute a threshold for the differential

error?

Answer:

For 51 km baseline, in differential mode the errors in positioning reduced from

100-200 m to up to 2 meters on vertical and horizontal component.
conservative bound

= 2000 m orbit error

b = 51000 m baseline length
= 20000000 m approximate Sat Rx distance
Result: < 5.1 m

TUTORIAL 3
Session A: Fixing DD ambiguities one at a time: UPC1 UPC2
A.1 Trying to fix ambiguities in Single Frequency
Questions (pg.22) :

Explain what is the meaning of this plot .

Is it possible to identify the integer ambiguity?

How reliability can be improved?

Answers:

This plot resembles the ambiguities for the L1 frequency measurements, e.g.
P1(code) and L1(carrier), which were used to compute the ambiguity term N1jk.
As it can be seen, the representation of ambiguities is dominated by noise.

It is not possible to identify the integer ambiguity because nothing is concentrated

around a certain value and the noise is too big.

Reliability can be improved by using another frequency or by using a code with

less noise.

Questions (pg.23) :

Explain what is the meaning of this plot .

Is it possible to identify the integer ambiguity?

How reliability can be improved?

8

Answers:

In this plot it has been represented the ambiguities for L2 frequency using
P2(code) and L2(carrier) measurements to compute the ambiguity term N2jk.

As in the previous case for L1, we cannot identify the integer ambiguity because
of the dominant noise. Nothing is concentrated around a value.

Reliability can be improved by using another frequency or by using a code with

less noise.

Question (pg.25) :

Discuss why the ambiguities cannot be fixed by rounding-off the expression

DDN1=[DDL1 DDP1] / 1 .

Answer:

The ambiguities cannot be fixed by rounding-off the expression

because the code used is too noisy. 50 cm of noise is more than 2 times the
wavelength of L1 (1=19 cm).
Question (pg.26) :

Discuss why the ambiguities cannot be fixed by rounding-off the expression

DDN1=[DDL2 DDP2] / 2 .

Answer:

The ambiguities cannot be fixed by rounding-off the expression

because the code used is too noisy. The noise is up to 1 m which is more than 2
times the wavelength of L2 (2=24.4 cm).
Question (pg.28) :

Discuss the plot. What is the level of the noise? Compare the noise with the
wavelength 1=19 cm.
9

Answer:

The noise is two order of magnitude smaller than the noise of the code. It can be
seen that the level of noise is at millimeter level and in comparison with the
wavelength for L1 frequency is significantly smaller, it is around 5 mm.

Question (pg.29) :

Discuss the plot. What is the level of the noise? Compare the noise with the
wavelength 1=24.4 cm.

Answer:

For L2 frequency is a similar situation as the noise is also at millimeter level

being significantly smaller than the wavelength of the second carrier.

A.2 Dual Frequency Ambiguity Fixing

Question (pg.34) :

DDNw = 3 ?.

Answer:

The ambiguity is not 3 in this case as DDNw depends on the code noise.

Question (pg.35) :

DDNw = 0 ?.

Answer:

Is the same case for PRN03, the ambiguity is not 0 for as DDNw depends on the
code noise.

Session B: Assessing the fixed ambiguities in navigation: Differential

positioning of UPC1-UPC2 receivers.

B.1 Repairing the DDL1 and DDL2 with the ambiguities fixed
Question (pg.63) :

Explain what is the meaning of this plot.

10

Answer:

It represents the carrier free of ambiguities and geometry and it can be seen the
noise of the carrier which is at millimeter level.

Question (pg.64) :

Explain what is the meaning of this plot.

Answer:

The plot represents the ambiguities together with the noise, as the double
difference term DDN1 has been ignored from the subtraction from the carrier
measurement.

Question (pg.65) :

Explain what is the meaning of this plot.

Why a trend and a discontinuity appears?

Answers:

The plot represents the variation of the geometric term DDRho over time

The trend is due to the geometric term and the discontinuities are caused by the
receivers clock adjustment.

Question (pg.67) :

Explain what is the meaning of this plot.

Answer:

This plot represents what is left after repairing the carrier for the ambiguities and
removing the geometric term, which is the carrier noise. The noise is at millimeter
level. We can see that the noise is slightly higher than on L1 and this is due to the
fact that L2 has less power.

11

Question (pg.68) :

Explain what is the meaning of this plot.

Answer:

By making the difference between DDL2 and DDRho we obtain the ambiguities
over time together with the noise.

Question (pg.69) :

Explain what is the meaning of this plot.

Why a trend and a discontinuity appears?

Answers:

In this plot we have the variation of geometry as the DDRho term was ignored.

The trend is due to the geometric term and the discontinuities are caused by the
receivers clock adjustment.

B.2 Assessing the fixed ambiguities in navigation

Question (pg.79) :

What is the expected accuracy when positioning with carrier after fixing the
ambiguities ?

Discuss why a trend and a discontinuity appears ?

Answers:

The expected accuracy using the carrier measurements which were repaired for
the ambiguities is centimeter level or better.

The trend is due to the synchronism errors between the receivers, and due to the
influence of geometry which was not used in obtaining this solution. The
discontinuities are caused by the receivers clock adjustment. To get good results
we must take into account the geometry or have good synchronism.

12

Question (pg.90) :

Discuss why the results have improved, achieving centimeter level navigation.

Answer:

The results improved because we have taken into account the double difference
geometric term in the estimation process, eliminating the geometry variation
achieving an increase in the accuracy. Because now we have the dominant term
taken into the estimation it reduced significantly the effect of the synchronism
error.

Question (pg.87) :

Compare the results with the previous ones computed from DDL1.

Answer:

Compared with the previous case, the results using second frequency have a little
more degradation on the up component but is still in the range of centimeter
accuracy.

Question (pg.93) :

Discuss the results. What is the effect of the wrong fix?

Answer:

The effect of the wrong ambiguity fix is the amplification of the error by one
order magnitude. Fixing wrong the ambiguities will affect significantly the final
position.

Session C: Fixing DD ambiguities with LAMBDA method

C.1 Fixing the DDN1ambiguities with LAMBDA Method

Questions (pg.105) :

Can the ambiguities be well fixed?

Is the test resolutive?

13

Compare the fixed ambiguities with those obtained in the previous exercises when
fixing the ambiguities one at a time. Are the same results found?

What is the elapsed time to needed fix the ambiguities? And in the previous
exercise when fixing the ambiguities one at a time?

The values found for the ambiguities are the same than in the previous case?

Answers:

Yes, the obtained ambiguities are well fixed.

In this case, yes. By looking at the ratio between the square norm of the two
candidates, it gives a big value (3.107). The lower the ratio is, the greater the risk
of fixing wrong the ambiguity is, which is not the case in the current situation.

By looking at the fixed ambiguities for L1 in the previous exercise (pg.55):

DDN1 = [2 1 2 -1 4 7 1 4]
And analyzing the new results using LAMBDA method we have:
afixed(: , 1) = [2 1 2 -1 4 7 1 4]
We can conclude that the results are the same.

For the LAMBDA method, the time lapsed in this case was 15 seconds between
the two epochs taken in consideration t1=18000 and t2=18015. In the previous
case the time interval chosen was 18000:19900, the seconds lapsed being 1900.

Yes, they have the same values:

DDN1 = [2 1 2 -1 4 7 1 4]
afixed(: , 1) = [2 1 2 -1 4 7 1 4]

Questions (pg.117) :

Can the ambiguities be well fixed?

Is the test resolutive?

Compare the fixed ambiguities with those obtained in the previous exercises when
fixing the ambiguities one at a time. Are the same results found?

What is the elapsed time to needed fix the ambiguities? And in the previous
exercise when fixing the ambiguities one at a time?

The values found for the ambiguities are the same than in the previous case?
14

Answers:

Yes, the obtained ambiguities are well fixed. An interesting fact, in the case of L2,
we have two ambiguities which were fixed correctly by rounding directly:
round(a)= [-1 -2 1 2 1 -2 3 -1]
afixed(: , 1)=[-1 1 -1 2 2 -1 1 0]

Analyzing the ration of the square norm of the two candidates (3.540) it can be
concluded that the results can be trusted.

In the previous case, the values of the fixed ambiguities for L2 are:
DDN2 = [-1 1 -1 2 2 -1 1 0]
And after using LAMBDA method:
afixed(: , 1) = [-1 1 -1 2 2 -1 1 0]
We can conclude that they are the same.

The time lapsed for LAMBDA method is 15 seconds between the 2 epochs used
in computations. And in previous case the time lapsed for the ambiguity fixing
was 1900 seconds [18000:19900]

Yes, they have the same values:

DDN2 = [-1 1 -1 2 2 -1 1 0]
afixed(: , 1) = [-1 1 -1 2 2 -1 1 0]

Questions (pg.125) :

Can the ambiguities be well fixed?

Is the test resolutive?

Compare the fixed ambiguities with those obtained in the previous exercises when
fixing the ambiguities one at a time. Are the same results found?

What is the elapsed time to needed fix the ambiguities? And in the previous
exercise when fixing the ambiguities one at a time?

The values found for the ambiguities are the same than in the previous case?

Answers:

No, the ambiguities cannot be well fixed neither by rounding the decorrelation
floated solution nor by decorrelation and integer LS search solution. In order to be

15

well fixed the receivers must be synchronized using the time-tagged reference
measurements or use the geometry term in computation.

The test does not give us a confident result. The ratio between the square norm of
the best and second best candidate is too small which mean they both are similar

No, the results are not the same in this case.

The time lapsed for LAMBDA method is 15 seconds between the 2 epochs used
in computations. And in the previous case the time lapsed was 1900 seconds.

No, they are not the same.

DDN1 = [2 1 2 -1 4 7 1 4]
afixed(: , 1) = [-1 -12 18 2 -4 8 9 4]

Questions (pg.133) :

Can the ambiguities be well fixed?

Is the test resolutive?

Compare the fixed ambiguities with those obtained in the previous exercises when
fixing the ambiguities one at a time. Are the same results found?

What is the elapsed time to needed fix the ambiguities? And in the previous
exercise when fixing the ambiguities one at a time?

The values found for the ambiguities are the same than in the previous case?

Answers:

No, the ambiguities cant be well fixed neither by rounding the decorrelation
floated solution nor by decorrelation and integer LS search solution. In order to be
well fixed the receivers must be synchronized using the time-tagged reference
measurements or use the geometry term in computation.

The test does not give us a confident result. The ratio between the square norm of
the best and second best candidate is too small which means they both are similar.

No, the results are not the same in this case.

The time lapsed for LAMBDA method is 15 seconds between the two epochs
used in computations. And in the previous case the time lapsed was 1900 seconds.

No, they are not the same.

DDN2 = [-1 1 -1 2 2 -1 1 0]
16

afixed(: , 1) = [-3 7 -6 13 -3 19 -3 24]

TUTORIAL 4
Session A: Differential positioning of IND2-IND3 receivers (baseline: 18 meters)
A.2 IND2-IND3 Baseline vector estimation with P1 code
Question (pg.35) :

What is the level of accuracy ?

Why does the solution degrade when taking only two epochs?

Answer:

The level of accuracy is at the tens of centimeters. In the case when only two
epochs were used the following errors were obtained:
East 0.85 m
North 0.62 m
Up 1.78 m

The solution degrades because less information have been used in the estimation
process and also the code noise is less mitigated.

A.3 IND2-IND3 Baseline vector estimation with L1 carrier

Question (pg.46) :

Explain what is the meaning of this plot.

Answer:

In this plot is represented the noise of the carrier over time.

Question (pg.47) :

Explain what is the meaning of this plot.

Answer:

17

If we correct the double difference carrier only for the geometric term DDRho,
then what is left are the ambiguities with the noise of the carrier.

Question (pg.48) :

Explain what is the meaning of this plot.

Answer:

In this plot is represented the evolution of geometry as the term DDRho has been
ignored from the subtraction.

Questions (pg.57) :

Discuss the results by comparing them with previous ones with DDL1 carrier.

Discuss the pattern seen in the plot.

Answer:

We can see that the errors using DDL1 carrier were up to 2 cm while in the case
of DDP1 is up to 2 m. This is due to the code noise.

The pattern follows the code noise and possible multipath effect.

A.4 IND2-IND3 Baseline vector estimation Geometry diversity effect

Questions (pg.61) :

Has the accuracy improved ?

Are the ambiguities well fixed ?

Has the reliability improved? Why ?

Answer:

Yes, the accuracy increased from tens of centimeter to centimeter level.

Yes, the ambiguities are well fixed.

The reliability improved. If we look at the ratio of the square norm of the two
candidates we can see that it has a very high value. This means that we can trust
the results.
a) sqnorm(2) / sqnorm(1) = 73.7727325628055.
b) Ambiguities fixed from the LS integer search
18

afixed(: , 1) = [-8 20 -9 -8 -10

0 -8]

c) Ambiguities fixed by rounding the decorrelated floated solution:

afix=iZ*round(az);
afix' = [-8 20 -9 -8 -10

0 -8]

d) Ambiguities fixed by rounding directly the floated solution:

round(a) = [-8 20 -9 -8 -10

0 -8]

A.5 IND2-IND3 differential positioning with P1 code

Question (pg.71) :

What is the level of accuracy ?

Why does the solution degrade when taking only two epochs?

Answer:

The level of accuracy is at meter level when dealing with code positioning.

If we compute the 3D error as

we can see when all the epochs

were used, the error is 60 cm and with only two epochs is 2 m. The noise
cannot be mitigated with only two epochs.

A.6 IND2-IND3 differential positioning with L1 carrier

Question (pg.79) :

Can the ambiguities be well fixed?

Has the reliability improved? Why?

The values found for the ambiguities are the same as in the previous case?

Answer:

Yes, the ambiguities can be fixed. If we analyze the ratio of the square norm of
the two candidates we can trust the results as it is big (4.433).

The reliability indeed improved because this time the geometry has been taken
into account in the estimation process.

Yes, the values found for the ambiguities are the same as in the previous case.
They do not change.

19

Question (pg.83) :

Is the accuracy similar to that in the previous case, when estimating the baseline
vector?

Answer:

When the baseline vector has been estimated with the L1 carrier, the following
errors were obtained:
x(1:3) bsl_enu = [-0.012745 -0.006427 0.0038664]
And now when using the computed differential position, the results are
x = [-0.012790 -0.0064170 0.0036900]
We can conclude that the results are similar.

Question (pg.87) :

Compare this plot with the obtained previously when estimating the baseline from
the time-tagged measurements. Are the errors similar?

Answer:

It can be identified the same error pattern and error magnitude. This is due to the
noise present in the L1 carrier.

Session B: Differential positioning of IND1-IND2 receivers (baseline: 7 meters,

but synchronization errors)
B.3 IND1-IND2 differential positioning with L1 carrier
Question (pg.102) :

Can the ambiguities be fixed?

Give a possible explanation about why the ambiguities cannot be fixed?

Answer:

No, the ambiguities cannot be fixed as the ratio of the square norm between the
two best candidates is too small.

20

Because of the IND1 receiver clock drift, synchronization errors will be caused
between IND1 and IND2 which will not make possible the ambiguity fixing in
two epochs separated by 15 seconds.

Question (pg.107) :

Can the ambiguities be fixed?

Give a possible explanation about why the ambiguities cannot be fixed?

Answer:

No, the ambiguities cannot be fixed as the ratio of the square norm between the
two best candidates is too small which makes them too similar.

Still the increase in the time interval will not facilitate the ambiguity fixing,
impediment caused by the synchronism error.

Question (pg.112) :

Have the results improved?

Has the reliability improved?

Why it is not possible to fix the ambiguities?

Answer:
NOTE: Cannot compute using the epoch t2=17000 as the measurements were
selected at the beginning in the time interval 14500:16500. So t2=16500 has been chosen

No, the results did not improve.

By analyzing the ratio of the square norm between the two candidates we can
conclude that the reliability did not improved.
a) sqnorm(2) / sqnorm(1) = 1.53311684476613.
b) Ambiguities fixed from the LS integer search
afixed(: , 1) = [8 -17 24 -12

9 10

8]

c) Ambiguities fixed by rounding the decorrelated floated solution:

afix=iZ*round(az);
afix' = [9 -15 21 -10

9
21

9]

d) Ambiguities fixed by rounding directly the floated solution:

round(a) = [8 -17 24 -12 10 10

9]

Is not possible because the clock drift of IND1 receiver is too large.

Question (pg.113) :

Discuss how the relative receiver clock offset can affect the baseline estimation.

Answer:

As it can be seen from the previous three tables, the IND1 receiver has a very
large clock drift in comparison with the other two. The large clock drift affects the
time-tagged measurements broadcasted by the reference station which degrade the
process of the estimation for the baseline having as reference station: IND1.

B.4 IND1-IND2 differential positioning with L1 carrier (using the computed differential
corrections)
Question (pg.123) :

Can the ambiguities be fixed now?Why?

Discuss why the synchronism errors affect the two differential positioning
implementations.

Answer:

Yes, the ambiguities can be well fixed now. The synchronization errors will not
be an issue as computed differential corrections have been used. Also, by looking
at the ratio of the square norm between the two candidates we can deduct that is a
good fixing. But the value of the ratio should be above 3 for a better certainty.
Moreover by rounding directly the floated solution we can identify 3 correct fixed
(-17, 22 , 17).

Although the synchronization errors is not the main issue when using computed
differential correction it still have an effect. And for the relative positioning the
synchronization error is the main contributor for the wrong ambiguity fixing. The
internal receiver large clock drift affect the time-tagged measurements and the
pseudorange corrections (in a less manner the latter ones).

22

Question (pg.127) :

Explain what is the meaning of this plot.

Answer:

In this plot is represented the noise of the carrier over time.

Question (pg.128) :

Explain what is the meaning of this plot.

Answer:

If we correct the double difference carrier only for the geometric term DDRho,
then what is left are the ambiguities with the noise of the carrier.

Question (pg.129) :

Explain what is the meaning of this plot.

Answer:

In this plot is represented the evolution of geometry as the term DDRho has been
ignored from the subtraction.

Question (pg.131) :

Is the accuracy similar to the previous case, when estimating the baseline vector?
Why?.

Answer:

The accuracy is not similar with the previous case, where it was around tens of
centimeter. Now, using computed differential corrections the accuracy is around
centimeter level. The synchronization error does not have a great impact on the
estimation when using computed differential corrections.

Question (pg.135) :

Discuss the accuracy achieved and the possible error sources that could affect his
result (e.g. Antenna Phase Centres).

Answer:
23

With differential corrections we have near 1 cm level accuracy. Possible error

sources can be the carrier noise, antenna phase center.

Question (pg.138) :

Discuss the results by comparing them with the previous ones with DDL1 carrier
in the relative positioning implementation.

Answer:

The results are noisier when using code measurements, the noise being much
higher thus the accuracy is at meter level.

Question (pg.142) :

Discuss why does the accuracy degrades respect to the previous case. Why this
large error appears?

Answer:

In relative positioning, when time-tagged measurements are being used, they are
affected by the large clock drift of the reference station. This translates in errors in
positioning as it can viewed in the plot.

Session C: Differential positioning of PLAN-GARR receivers (baseline: 15 km,

Night time): tropospheric effects

C.2 PLAN-GARR differential positioning with L1 carrier (using the computed differential
corrections)
Questions (pg.155) :

Can the ambiguities be fixed with a certain degree of confidence?

Repeat the computations taking: t= 14500 and 14900.

Repeat the comoutations taking: t= 14500 and 15900.

Discuss why the ambiguities cannot be fixed.

Answers:

The ambiguities cannot be fixed with a certain degree of confidence.

24

sqnorm(2) / sqnorm(1) = 1.018799 ambiguities cannot be fixed.

sqnorm(2) / sqnorm(1) = 1.118400 ambiguities cannot be fixed.

The ambiguities cannot be fixed because the ionosphere and tropospheric effects
were not taken into account. For a baseline of 15 km the ionosphere and
troposphere do not cancel in double difference.

C.3 PLAN-GARR differential positioning with L1 carrier (using the computed differential
corrections including troposphere)
Questions (pg.163) :

Can the ambiguities be fixed now?Discuss, why?

Answers:

Yes, the ambiguities can be fixed now. We have taken into account the
differential troposphere term. In this case, neglecting the ionosphere will not
produce significant harm to the results as the observations were made during night
time when the activity of the ionosphere is lower.

Question (pg.167) :

Explain what is the meaning of this plot.

Answer:

In this plot it can be observed the tropospheric delay together with the noise of the
L1 carrier.

Question (pg.168) :

Explain what is the meaning of this plot.

Answer:

The plot resembles the representation of the noise of the L1 carrier and one can
observe a tendency due to the slight ionosphere. But it is small as the
measurements were made during the night.

25

Question (pg.169) :

Explain what is the meaning of this plot.

Answer:

It represents a zoomed version of the previous plot where the noise is more clear.

Question (pg.170) :

Explain what is the meaning of this plot.

Answer:

It represents a superposition of all the effects. The big variation is due to the
geometric term DDRho which is dominant.

Question (pg.176) :

Discuss the possible sources of the bias found in the vertical component.

Answer:

In this estimation process the wet component has not been taken into account and
it represents 10% of the tropospheric delay. The bias can also be related to the
antenna phase center of the receivers.

Question (pg.180) :

Discuss the possible sources of the bias found in the vertical component.

Answer:

The error caused by the antenna phase center has been corrected and it can be
observe a slight improvement in the solution. It still remains the effect of the wet
component of the troposphere together with the slight influence of the ionosphere.

Question (pg.183) :

Discuss the results by comparing them with the previous ones with DDL1 carrier.

Answer:

Accuracy of the code is at meter level. It is not possible to distinguish the effects
of the other error sources because everything is under the level of the noise.
26

Session D: Differential positioning of PLAN-GARR receivers (baseline: 15 km,

Day time): tropospheric and ionospheric effects

D.3 PLAN-GARR differential positioning with L1 carrier (using the computed differential
corrections including troposphere)

Question (pg.196) :

Explain what is the meaning of this plot.

Answer:

In this plot it can be observed the tropospheric delay together with the ionosphere
and the noise of the L1 carrier.

Question (pg.197) :

Explain what is the meaning of this plot.

Answer:

The plot resembles the representation of the noise of the L1 carrier and one can
observe a tendency due to the ionosphere.

Question (pg.198) :

Explain what is the meaning of this plot.

Answer:

It represents a zoomed version of the previous plot where the noise and the trend
of the ionosphere are more clear.

Question (pg.199) :

Explain what is the meaning of this plot.

Answer:

It represents a superposition of all the effects. The big variation is due to the
geometric term DDRho which is dominant.
27

Question (pg.205) :

Discuss on the remaining error sources which could explain the error found in
the North, East and Vertical components.

Answer:

The positioning solution is influenced by not taking into account the differential
ionospheric term in the estimation process together with ignoring the wet
component of the troposphere.
D.4 PLAN-GARR differential positioning with L2 carrier (using the computed differential
corrections including troposphere)
Question (pg.214) :

Explain what is the meaning of this plot.

Answer:

In this plot it can be observed the tropospheric delay together with the ionosphere
and the noise of the L2 carrier.

Question (pg.215) :

Explain what is the meaning of this plot.

Answer:

The plot resembles the representation of the noise of the L1 carrier and one can
observe a tendency due to the ionosphere.

Question (pg.216) :

Explain what is the meaning of this plot.

Answer:

It represents a zoomed version of the previous plot where the noise and the trend
of the ionosphere are more clear.

Question (pg.217) :

Explain what is the meaning of this plot.

28

Answer:

It represents a superposition of all the effects. The big variation is due to the
geometric term DDRho which is dominant.

Question (pg.223) :

Discuss the possible sources of the bias found in the vertical component.

Answer:

The possible main bias found in the vertical component appears by not taking into
account the ionospheric delay together with ignoring the wet component of the
troposphere.

D.5 PLAN-GARR differential positioning with L2 carrier (using the computed differential
corrections including troposphere and Klobuchar iono)
Question (pg.225) :

Discuss this plot.

Answer:

In this plot are presented the values for the ionospheric model using Klobuchars
algorithm which is not perfect.

Question (pg.226) :

Why a larger dispersion is found at a low elevation.

Answer:

The ionospheric delay is elevation dependent. At low elevations, the delay will be
higher as the signal has to travel through a bigger part of the ionospheres layer
while on high elevation it travels through less.

Question (pg.228) :

Discuss the noise seen in the plot.

29

Answer:

The Klobuchar model mitigates about 50% of the ionospheric delay, it does not
completely remove it. So, in the plot one can see a pattern due to missmodeling
together with the carrier noise.

Question (pg.232) :

Discuss the results.

Answer:

The navigation solution has a slight improvement in the accuracy.

D.6 Unambigous DDSTEC determination
Question (pg.234) :

Discuss the noise seen in the plot.

Answer:

The noise is due to the carriers but the trend is caused by the ionosphere as the
DDSTEC determination is not perfect but is much better than Klobuchars.

Question (pg.235) :

Why do we have an elevation-dependent pattern ?

Answer:

It has not been taken into account that the L1, L2 carries arrive under different
angles in APC of the receivers and thats why we have elevation dependency.

Question (pg.237) :

Discuss the noise seen in the plot.

Answer:

The present noise is due to the carrier noise as the differential terms: L1, , N,
Troposphere, Ionosphere were taken into account.

30

Question (pg.239) :

Why do we have an elevation-dependent pattern ?

Answer:

It has not been taken into account that the L1 carrie arrive under different angles
in APC of the receivers and thats why we have elevation dependency.

Question (pg.243) :

Is any bias expected due to the L1-LC APCs, when removing the ionosphere using
the unambiguous DDSTEC ?

Answer:

Yes, the L1-LC APC must be taken into account in the modeling process.
Question (pg.247) :

Compare this iono-free solution with that obtained with DDL1, removing the
troposphere and ionosphere using the unambiguous DDSTEC. Are the results the
same ? Why ?

Answer:

The iono-free solution has the same pattern as the one obtained with DDL1.
Processing with iono-free combination (LC) is the same as double differencing
with the carriers.

31