Professional Documents
Culture Documents
I NTRODUCTION
Nowadays more and more mobile users are present also in indoor environments. Therefore it is crucial that mobile communication systems cover also indoor users. This can be performed in two ways, having a base station (BS) indoors or having the BS outdoors and letting the signal to propagate through
the building walls. This is defined as the outdoor-to-indoor scenario [1].
The most important parameter in practically every channel
modeling work, is the analysis for the propagation loss (path
loss) in a certain receiver (Rx) location. Path loss is determined
as the sign inverse of the received signal power. In the COST
231 outdoor-to-indoor model, path loss consists of three different sections, namely outdoor, indoor and through-wall parts.
The proposed extension will give more accuracy in every section. Furthermore, the indoor propagation loss section is simplified to allow for more extensive use of performance analysis.
Proposed model can be applied also in indoor-to-outdoor scenario where, for example, a user is sitting in the garden and
using a laptop which is connected to the internet via indoor BS
[2].
The COST 231 [3] presented a model for predicting the
propagation path loss in outdoor-to-indoor scenario based on
the above mentioned three propagation parts. Applied frequency range was 900-1800 MHz which is certainly too low
for beyond 3G systems. For this reason alone, extension of the
COST 231 model to more up-to-date frequencies is needed.
Later in 2002, Miura et al. [4], extended the COST 231 model
to a frequency level of 8.45 GHz. In addition Miura presented
some modifications to the COST 231 model that take into account the wall openings such as doors and windows. Clearly,
Miura model includes indoor angle description relative to the
wall openings which is too detailed knowledge for general
outdoor-to-indoor simulations. Oestges et al. presented in [5]
an extension to the COST 231 model in a frequency of 2500
MHz. This model improved the accuracy of the indoor propagation loss part by introducing the impact that the height of the
Rx antenna has on the path loss. This is, however, again relatively detailed information about the Rx antenna location and a
simpler model is required.
Namely the following modifications to the COST 231 model
are proposed:
Outdoor propagation loss section is adjusted to be more
accurate than in COST 231 model where it is obtained
simply by free space loss.
Parameters We and W Ge will be derived separately for
3.7 and 5.25 GHz directly from measurements. Values
will be given to glass walls as well concrete walls.
Indoor propagation loss in COST 231 is calculated in a
way that does not allow extensive use of performance simulations. We propose a simpler, yet equally accurate, formula for calculating the indoor propagation loss.
The proposed extension can be applied also in indoor-tooutdoor scenario. Measurements related to the parameter determination are performed within IST-WINNER project. This
paper is organized as follows. In Section II, the COST 231
channel model is briefly presented. Section III describes the
parameter derivation and the measurements performed to obtain the parameters. Section IV shows the path loss results of
the measurements conducted at the University of Oulu and their
comparison to the proposed model. Finally, Section V draws
the conclusions.
II
COST231 MODEL
(1)
(2a)
2 = (d 2)(1
D 2
) .
S
(2b)
Ltw = We +W Ge (1
D 2
) = We +W Ge (1sin())2 , (3)
S
D ERIVATION
OF
Frequency [GHz]
Bandwidth (B) [MHz]
Length of the code [chips]
Used antenna configuration Nt Nr
Transmitting power [dBm]
PARAMETERS
Window
Door
6.7m
Rx
5.7m
1m
2.8m
4m
f [GHz]
), Lf ), (4)
5
where Lout and Lin are given in (4) and (5), respectively. Of
course the optimal accuracy is obtained by varying the We and
W Ge values according to the wall material next the location of
the indoor Rx. However, for performance simulations this is
too detailed information and, therefore, a compromise is derived by choosing We and W Ge values to represent typical
modern office building in which external walls include concrete as well as glass.
25
20
15
10
10
20
30
40
50
60
Measurement angle [deg]
70
80
90
3.7 GHz
We = 5
W Ge = 14
We = 20
W Ge = 14
5.25 GHz
We = 6
W Ge = 15
We = 22
W Ge = 26
(5)
regardless the location of the Rx. This can be done by assuming that there exists one internal wall per 10 meters distance.
In the IST-WINNER [1] indoor radio channel measurements
conducted in a typical office environment in the University of
Oulu, the loss per wall was found to be approximately 5 dB
[8]. By selecting to be 0.5 dB/m, the indoor propagation
loss is accurate regardless of the presence of the internal walls.
d is the distance from the external wall to the indoor Rx. Frequency dependency is discarded in indoor propagation since
the distances are typically short compared to outdoor distances.
Furthermore, the impact of indoor propagation loss is relatively
small compared to the Lout and Ltw .
IV
R ESULTS
Based on these parameters, outdoor-to-indoor and indoorto-outdoor propagation loss is approximated and compared to
measured propagation loss values obtained from measurements
performed in the University of Oulu in 2006 and 2007. The settings for the measurements are given in Table 1. In 5.25 GHz
outdoor-to-indoor measurements, BS was located in two different spots and the mobile indoor antenna was mounted in a trolley and it was moved inside the building in different floors and
along different wall materials including concrete, windows and
plaster board. The height of the Rx antenna was 1.5 meters and
the height of the outdoor antenna was 6 meters. In indoor-tooutdoor scenarios, the measurements were conducted in three
different indoor base station locations. The height of the BS antenna was 2 meters. Some of the routes traveled as well as the
BS locations in outdoor-to-indoor and indoor-to-outdoor scenarios are shown in Figure 4. In 3.7 GHz only part of the routes
were traveled while the BS was located in BS1 spot in Figure
4. Hence, the measured path loss values with 5.25 GHz cannot
be compared with 3.7 GHz.
Figures 5 and 6 show the measured propagation loss in
outdoor-to-indoor scenario compared to the modeled path loss
values given by the proposed model and by the original COST
231 model in 5.25 GHz frequency. It can be seen that the COST
231 model underestimates the path loss and the proposed extension is significantly more accurate. Indoor-to-outdoor measurement results could not be provided in this paper.
V
C ONCLUSION
The parameters for the proposed outdoor-to-indoor propagation loss model are given in Table 3. The path loss formula
reads as
In the COST 231, the outdoor propagation loss part is obtained simply with free space loss. We propose to use
WINNER urban micro cellular model given in (5).
(6)
125
Measured path loss
Proposed model
COST231
120
115
110
105
100
95
90
85
80
75
70
0
Figure 4: Routes traveled in outdoor-to-indoor and indoor-tooutdoor scenarios with 5.25 GHz. Red and green marks show
the BS locations in outdoor-to-indoor and indoor-to-outdoor
scenarios, respectively.
130
Measured path loss
Proposed model
COST231
120
0.4
0.6
Normalized snapshot
0.8
Figure 6: Measured and modeled path loss values in outdoorto-indoor scenario with 5.25 GHz.
results show that the proposed extension is significantly more
accurate than the original COST 231 model and should be
adopted for propagation loss prediction model in outdoor-toindoor and indoor-to-outdoor scenarios.
ACKNOWLEDGMENT
This work has been performed in the framework of the IST
project IST-4-027756 WINNER II, which is partly funded by
the European Union. The work of Mikko Alatossava was supported by Tekniikan edistamissaa tio, Kaupallisten ja teknillisten tieteiden tukisaa tio and Oulun yliopiston tukisaa tio.
110
100
90
R EFERENCES
[1] P. Kyosti, IST-4-027756 WINNER II D1.1.1 v.1.0, WINNER II Interim
Channel Models, IST-WINNER II, Tech. Rep., 2006.
80
70
0
0.2
0.2
0.4
0.6
Normalized snapshot
0.8
Figure 5: Measured and modeled path loss values in outdoorto-indoor scenario with 3.7 GHz.
COST 231 the parameters are for lower frequencies and
the measurements indicated that the COST 231 values are
not sufficient for higher frequencies.
Indoor propagation loss in the original COST 231 is calculated in (2) depending on the location of indoor Rx in
a corridor or in a room. We propose a simple formula of
Lin = d, where is the loss in dB per meter and d is the
internal distance in meters. This is to allow for more extensive use of performance simulations, because the user
does not need information about whether the Rx is inside
a room or in a corridor.
Measurements to derive the parameters for each of the three
propagation sections are performed within IST-WINNER. The