Sei sulla pagina 1di 6

1.1.

How to approximate Atolls Standard Propagation Models

1.1.1.1

Methodology

Atolls Standard Propagation Models can be approximated in Mentum Planet


using the Planet General Model. Although perfect replication of the models are not
always possible, due to some differences in terms of propagation modelling, predictions
have proved to be similar with both propagation models.
Here are the steps to be followed to create a PGM model, based on a given
Standard Propagation Model:
-

Create a new PGM model.

Set the same frequency as the one defined in the Standard Propagation Model.

Set the same Rx height as the one defined in the Standard Propagation Model.

K1 and K2 factors: Atoll models use separate K1 and K2 factors for the near the
transmitter and far from the transmitted; a distance threshold needs to be input to
define when the near factors should be used, and where the far factors should be
used. This can be modelled with the Planet General Model as well. In the Planet
General Model parameters, select the 2 Piece type, and input the distance that was
defined in the SPM model.
The SPM models also has the ability to specify different K1 and K2 factors for Line of
Sight and Non Line of Sight areas (for both near and far distances); the PGM
model does not offer such capability; it is hence recommended to use the NLOS K1
and K2 factors, as it will help approximate the Standard Propagation Model where it
matters more (i.e. Non Line of Sight areas).

The K3 value defined in the SPM model should be input as -K3 in the PGM model
(for instance, 5.83 should be input as -5.83).

K4 factors are equivalent in the SPM and the PGM models.

The K3 factor defined in the SPM model should be input as -K5 in the PGM model
(for instance, -6.55 should be input as 6.55).

K6 factors are equivalent in the SPM and the PGM models.

The Kclutter factor is used to multiply the per-clutter losses, called CAL (Clutter
Absorption Loss) values in Mentum Planet. If Kclutter is different from 1, then in
the PGM parameters, in the Path Clutter tab, check the enable path clutter option,
and:
o

Set the distance as defined in the Maximum distance cell in the SPM model.
If equal to 0, only the loss of the clutter class of the receiving Bin will be
accounted for.

Mentum confidential proprietary

Set the Function coefficient to the Kclutter values defined in the SPM
model.

Set the Function based on the option selected in the SPM model (Uniform in
the SPM model would be rectangular in the Planet General Model for
instance; this option gives the same weighting for all the clutter losses; the
other options in the Planet General Model are triangular, exponential and
logarithmic).

Figure 19: Planet General Model Path Clutter tab

If the Hilly terrain correction is used (i.e. if it does not say 0 No in the SPM
model), which was not the case in the Vancouver (and hence the Calgary and
Edmonton) models, further investigation of the SPM model would be needed to
analyse which correction factors should be enabled in the Planet General Model. It is
understood that this setting only impacts the predictions generated by the SPM
model for the Bins/pixels that are in line of sight. For more information on the various
correction factors in the PGM model, please refer to the Planet General Model
technical notes.

Predictions generated by the SPM model can be radial-based or Bin-based; in


Mentum Planet, all the predictions are radial-based, because predictions with Binbased methods are much slower, and studies have shown that radial-based
predictions are as accurate as Bin-based predictions; more information on this can
be provided if needed.

The grid calculations can be centred, in which case the predictions are done at the
centre of the Bins, or bottom left, in which case the predictions are done for the
bottom left of the Bins, with the SPM model. With the PGM model, the predictions
are always generated at the centre of the Bins.

Mentum confidential proprietary

If the Consider heights in diffraction option is set to 0 No in the SPM model


(which was the case for the models in Vancouver, then set all the clutter heights to
0m in the Planet General Model. If the Consider heights in diffraction option is not
set to 0 No, then assign the same clutter heights as the ones defined in the SPM
model (clutter tab). The clearance values defined in the SPM model are equivalent
to the clutter separation value defined in Planet.

With regards to the Effective Antenna Height options, the SPM and the PGM
models both offer various methodologies; after running multiple tests, the ground
reflection slope methodology seems to give better approximations of the SPM
predictions, regardless of the option selected in the SPM model. This was verified for
the both models in Vancouver, where the SPM dense urban model used the Height
above average profile option, while the suburban model was assigned the
enhanced slope at receiver option.

As mentioned earlier in the document, the PGM model uses a modified version of the
Epstein-Peterson algorithm to compute diffraction calculations; the SPM model offers
four choices (Epstein-Peterson, Deygout, Deygout with correction and Millington).
When approximating the SPM models in Planet, some differences may be noticed.
The differences will vary depending on the environments.

If the Receiver on top of clutter option is enabled (i.e. the parameter is not set to 0
No), the calculations (in Atoll) will assume that the receiver height is the sum of
the clutter height and the receiver height defined in the receiver tab (e.g. 20 + 1.5 =
21.5m). If this option is enabled, separate receiver heights should be set in the
PGMs .cpa file. The receiver heights should be equal to the clutter heights the
receiver height defined in the SPM model (e.g. 20 + 1.5 = 21.5m).

Limitation to free space loss: this is a feature offered by the SPM model (whereby the
pathloss will never be smaller than the free space loss) that is not available in the
PGM model.

In the SPM model, it is also possible to set different receiver heights for different
clutter classes; if this is the case, the same values should be input in the receiver
height column of the PGMs .cpa file.

1.1.1.2

SPM model approximation - verification

The verification of the approximation of the SPM models was done based on
statistics, so only statistical comparisons could be performed:

Mentum confidential proprietary

The statistics between the measurements and the approximated predictions in


Planet were computed; these statistics were compared against the statistics shown
in the model tuning report coming from Atoll. As shown on the table below, the
statistics between the measurements and the SPM model on the one hand, and the
statistics between the measurements and the approximated SPM model on the other
hand are almost identical for the dense urban model; for the suburban model, the
difference seems to be a bit more significant. In both cases, these statistics suggest
a good approximation of the SPM models in Mentum Planet.

Calibration Survey

Measurements Environment

Broughton
Georgia W
Georgia O
Global

13272
5770
3783
22825

DU
DU
DU
DU

Farrow
David Gray
Quality Inn
Global

16225
8648
21347
46220

SU
SU
SU
SU

Measurements

Environment

12592
20406

DU
SU

Validation Survey
Best Western
Johnson

SPM-Atoll
Mean
STDEV
1.5
5.5
-1.9
6.4
-2.5
6.7
0.0
5.9
0.3
-2.5
0.4
-0.2

6.0
4.4
6.9
6.1

SPM-Atoll
Mean
STDEV
0.6
5.8
0.6
4.9

SPM-Approximated
Mean
STDEV
1.2
5.7
-1.2
6.3
-1.9
7.0
0.1
6.0
0.1
-2.9
-1.9
-1.4

7.2
6.2
7.7
7.2

SPM-Approximated
Mean
STDEV
-1.0
6.5
-0.9
4.9

Table 5: SPM and SPM-Approximated predictions compared against measurements

The predicted values of the SPM models were given to Mentum, for the Bins with
measurements only, and the approximated predictions in Planet were compared
against the predicted values in Atoll. Again, these statistics suggest a really good
approximation of the SPM models.

Mentum confidential proprietary

Calibration Survey

"SPM Predictions" - "PGM-Atoll Predictions"


Measurements Environment
Mean
STDEV

Correlation

Broughton
Georgia W
Georgia O
Global

13272
5770
3783
22825

DU
DU
DU
DU

-0.2
0.6
0.5
0.2

2.6
1.9
1.8
2.3

0.97
0.99
0.99
0.98

Farrow
David Gray
Quality Inn
Global

16225
8648
21347
46220

SU
SU
SU
SU

-0.2
-0.4
-2.3
-1.2

3.2
3.6
3.1
3.2

0.96
0.98
0.97
0.97

Validation Survey
Measurements Environment
Mean
STDEV Correlation
Best Western
12592
DU
-1.6
2.9
0.97
JohnsonTable 6: Difference between
20406 SPM predictions
SU and Approximated
-1.5
2.6predictions
0.97
SPM

Here is also the distribution (pdf and cdf) of the difference between the predictions
generated with the SPM model in Atoll, and the predictions generated in Mentum planet
with the approximated SPM model, for the Bins that had measurements, like the two
previous tables, this suggests a really good replication of the SPM models, as the
difference observed are quite small when compared to the accuracy of the models:

Mentum confidential proprietary

Figure 20: Distribution of the difference between SPM and SPM-Approximated predictions

Mentum confidential proprietary

Potrebbero piacerti anche