Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Modeling new indoor and outdoor propagation models for

WLAN
Agustin Zaballos, Guiomar Corral, Albert Carn, Joan Lluis Pijoan
Enginyeria i Arquitectura La Salle, Universitat Ramon Llull, Barcelona Spain, EUROPE
Paseo Bonanova, 8, 08022 Barcelona Tlf: +34 93 2902400, Fax: +34 93 2902416
E-mail: {zaballos, guiomar, st05121, joanp }@salleURL.edu
Abstract

buildings such as university buildings, student halls, airports,


offices and so on.

This paper presents the modeling of new WLAN models for


indoor and outdoor environments. Based on the standard Opnet
models for WLAN nodes, the propagation loss estimation for
these two types of environment has been improved.

This paper proposes different models in order to obtain a better


model of the losses in this kind of environments. This is because
the standard WLAN model that comes with OPNET only uses
the free space propagation model. These new introduced models
are based on empirical models which let us get the path loss with
low computational time although the accuracy is lower than the
other models [1], [2], [3].

Different proposals are given in order to obtain a better indoor


coverage model as in university buildings, student halls, airports
and offices. The studied indoor models are Linear, One Slope,
Modified Free Space, Motley-Keenan and COST Multi-Wall.
The new outdoor model introduced is the CCIR Urban Model.
The scenarios used to simulate these new propagation models
are shown. Finally, results, conclusions and further work are
given.

Different objectives of the simulation experiments have been


defined. First of all, different WLAN networks have been
created to reflect their operation in the different topologies, like
Ad-hoc networks, IBSS and so on. Secondly, a simple network
with only two nodes has been created to test different
propagation models, not only indoor but also outdoor models.
These propagation models have been included in our new
WLAN model, changing the calculation of the received power in
the Transceiver Pipeline stage and adding new attributes to
different nodes. Thirdly, the new created models have been
simulated with their own environment; that is to say, a building
map has been introduced in the indoor models to better fit the
different parameters. So a better comparison can be made
between the different models. Finally, our university campus has
been modeled with OPNET to analyze its WLAN networks and
obtain a more accurate model of our real networks.

I. Introduction
Over the last few years, WLANs have gained strong popularity
in a number of vertical markets, including health-care, retail,
manufacturing, warehousing, and academic areas. These
industries have profited from the productivity gains of using
hand-held terminals and notebook computers to transmit realtime information to centralized hosts for data processing. Today
WLANs are becoming more widely recognized as a generalpurpose connectivity alternative for a broad range of business
customers.
Many wireless network standards have appeared up to now. The
most known standards belong to the IEEE 802.11 family, which
includes the popular 802.11b, the 802.11a and the 802.11g.
Other standards, such as HIPERLAN and HIPERLAN/2, had
also some importance but they didnt find as much acceptance in
market as the others [4].

The remaining sections in this paper are as follows: section II


describes the different models that have been evaluated; section
III focuses on the way the new models have been implemented;
sections IV and V explain simulation environment and results. In
conclusion, we present some suggestions for further work.

Two groups of standards can be distinguished in terms of the


frequency band they use. The first group includes those
standards that use the free 24 GHz band. The second group
includes those that use the other free band at 5 GHz. Another
classification of standards can be made according to the wireless
access technology they use. We can distinguish the standards
that define the CDMA DS/FH technology and the ones that use
an improved technology called OFDM. The systems based on
the latest technology can achieve upper data rates (up to 54
Mbps compared to the 11Mbps of the earlier technology).

II. Problem analysis


Predicting the propagation characteristics between two antennas
inside a building is important especially for the design of
cordless telephones and WLANs (Wireless Local Area
Networks). Also the design of cellular systems with indoor base
stations involves the use of indoor propagation models.
The indoor propagation channel differs considerably from the
outdoor one. The distance between transmitter and receiver is
shorter due to high attenuation caused by the internal walls and
furniture and often also because of the lower transmitter power.
The short distance implies shorter delay of echoes and
consequently a lower delay spread. As is the case in outdoor
systems, there are several important propagation parameters to

If we focus on the signals of wireless LANs, we have to


consider the propagation environments we will run into.
Basically, WLAN networks are the access part of wired
networks (trunk networks) and these are normally found inside
1

be predicted. The path loss and the statistical characteristics of


the received signal envelope are the most important for coverage
planning applications.
It is rather difficult to consider a new publication on mobile
radio as an actual novel book [1], due to the proliferation of
literature and contributions on this subject. Nevertheless, the
turbulence of the environment and the fast development of
mobile communications world tend to make any publication
rapidly obsolete and subsequently a dissemination of up-to-date
information is required.
Most of the models proposed in this paper have been obtained
from the COST 231 project [1]. This literature has been chosen
because it is an excellent synthesis of the fresh achievements
resulting from activities carried out during the past ten years by a
huge number of European experts in different areas of wireless
communications.
The considered propagation models in the literature are divided
into two groups: empirical models and deterministic models.
Earlier models were expressed in form of simple mathematical
equations which give the path loss as the output. The equations
were obtained by fitting the model measurements results (at
1800 MHz). Latter models are calculation methods which
physically simulate the propagation of radio waves [1], [2], [3].

Figure 1: Radio Pipeline Stages


We introduced six new propagation models; five indoor models
and another outdoor propagation model apart from the Free
Space model which is in the standard models. The indoor
models we chose, from the simplest to the more accurate one,
are: Linear, One Slope, Modified Free Space, Motley-Keenan
and COST Multi-Wall [1], [3]. The new outdoor model
introduced is the CCIR Urban Model [2]. Although the major
deployment in wireless is in indoor environments, sometimes the
outdoor coverage is also very useful. That is why not only
indoor but also outdoor models have been analyzed.

Between these two groups, the empirical ones minimize the


computation time in order to give results even if their accuracy is
low. The deterministic ones have a good accuracy but the
computation time needed to get the results is very high. It is for
this reason that we have adopted the empirical models to
introduce them in the WLAN models.
III. Modeling methodologies
In this section we describe the process we have followed to
introduce these new propagation models in the new WLAN
models. As stated before, all the propagation models
implemented in the WLAN nodes are empirical and they use
equations to calculate the path losses.

After programming all the models, it was necessary to introduce


the new parameters from each model into the new WLAN nodes
so that the users could adapt the parameters from the
environment where each node is. This was achieved by using the
node editor programmable environment. Due to the fact that
many parameters may have a range of possible values, this range
was limited and a default optimal value for each parameter was
used [1], [2]. The meaning of each parameter was described in
order to understand what each parameter does.

It is necessary to take two steps to achieve this goal. The first


one consists of knowing where the new algorithms of the new
empirical models must be introduced. OPNET provides different
type of nodes and links in the Radio module. The OPNET link
architecture is called Transceiver Pipeline. Its main function
consists in identifying when a packet arrives at the receptor and
realize if the transmission was correct. The link is modeled in
different stages. Every stage controls a basic parameter like
interferences, the signal/noise relation, the propagation delay and
so on.
Out of the 14 stages that make up the radio Transceiver Pipeline
[6],[7], one must be modified. The stage in which the path loss is
calculated is the seventh stage received power stage. This is
because these models calculate the path loss depending on the
simulation environment. In order to know the received power it
is necessary to be aware of the losses the path has introduced.
Thus a better model can be designed.

Figure 2: New attributes for the WLAN nodes

The new attributes introduced in the WLAN nodes are useful to


select the model for the simulation. Also the parameters
associated with every model can be modified with this
improvement.

Apart from seeing the differences in the path loss, a further step
was taken to demonstrate how the distance also affects the
parameter. To achieve the attempted goal a trajectory was
attached to the mobile node to move further and closer from the
mobile station.

Figure 3: New models


IV. Simulation test-bed
After modifying the WLAN nodes, it was necessary to test if
they worked well. The test-bed was implemented by using
OPNET 9.1 [6], [7].
Two different projects with many scenarios were created. The
first simulation environment was a simple network based on two
nodes, one acting as a base station and the other one as a mobile
node. Both are WLAN modified nodes as previously described
in the section above.

Figure 5: SNR comparison for the indoor models


The parameters used to get these results are listed in the
following table:
Model
One Slope
Linear
Modified
Free Space
Motley-Keenan

Figure 4: Test-bed Scenario


To test the five indoor propagation models, the distance between
the two nodes was set at 75 meters, which is in the coverage of a
typical Access Point. It is very difficult to compare the different
indoor propagation models because every model has different
associated parameters, like the number of walls, windows and so
on. This is why the different configurations have been randomly
established. However the models have to be checked to verify its
correct programming. The graphs and statistics can help to
verify the models.

Cost- Multi Wall

Parameters
Lo = 0 dB
n=4
= 05 dB/m
Lc = 1 dB
n = 2.0 (NLOS)
Kw = 3
Lw = 5 dB
Lc = 1 dB
Kw1 = 1
Kw2 = 2
Kf = 0
Lw1 = 3.4 dB
Lw2 = 6.9 dB
Lf = 18.3 dB
b = 0.46
Lc = 1 dB

Table 1: Values of the models parameters


As we can see in Figure 5, there is a significant difference
between results. Such difference could lead us to the belief that
it is not correct due to the fact that the differences are very large.
The new models must be analyzed in order to find out why such
differences take place.

The statistic that best shows the performance of the new WLAN
models is the SNR (Signal to Noise Ratio). From that statistic
we may observe the differences on the path loss calculation of
each model.
3

One of the reasons of these differences is the parameters set up.


Depending on the value of the parameters of each model the
results may vary considerably. These parameters have been
chosen from the optimal values found in the literature [1],[2].
Literature presents a range of values for each parameter so
values have been chosen to try to compare all the models. The
way they calculate the propagation loss is very different from
each other and it makes comparison among them very difficult.

The other scenario tests the performance of the new outdoor


propagation model, CCIR model [3], in comparison to the Free
Space model. In this case a distance between nodes of 2 Km has
been chosen. Such large distance has been used to point out how
it affects to the SNR results.
The SNR comparison between the Free Space and the CCIR
urban model shows how a 10% of building percentage in the
CCIR model reduces the SNR and how the trajectory of the
mobile node makes the SNR vary throughout the simulation
(Figure 6).

If we pay attention to Figure 5 we notice the following results.


The best SNR is achieved by one-slope model (ISM). That is
because it is a very simple model that does not take the
environment much into account. That makes the propagation
loss lower.

V. WLAN scenarios
As earlier stated, it is almost impossible to compare all the
indoor models due to the fact that they use different parameters
to adjust their results. These parameters take into account
different aspects of the propagation. Thus, they cannot be
directly compared.

The next two best SNR belong to motley-keenan (MK) and cost
multi-wall(MWM) models. These are very similar in the sense
that they calculate the propagation loss in a similar way. They
give the path loss as the free space loss added with losses
introduced by the walls and the latest also by the floors
penetrated by the direct path between the transmitter and the
receiver.

However, although direct comparison is not possible, the models


can be put in context to compare their results in a real
environment. This allows the parameters to be adjusted in a
more accurate way. In order to do this, some buildings from La
Salle campus have been chosen as a real environment to test the
models.

The SNR from linear attenuation model (LAM) gives a different


result from the others in terms of the graph line. This is because
the model assumes that the excess path loss (dB) added to the
free space loss is linearly dependent on the distance. This double
dependence on the distance makes the graph sharper.

A second project has been created in order to adjust the


particular parameters of each model from getting information of
the referenced map. Models like Motley-Keenan and Cost Multi
Wall allow us to adjust their parameters better because they
depend on how many walls and floors the signal has to penetrate
and what is the loss of each type of wall or floor. This
information can be found in the map by counting the walls
between a particular node and the WLAN server and assigning it
to the particular parameter. Two types of walls may be
distinguished, the thin ones and the thick ones (for Cost Multi
Wall model) as may be observed in the map.

The worst SNR belongs to modified free space model (MF). This
model computes the path loss similar to the free space loss with
adaptable exponent and offset. Herewith it assumes that the
excess path loss (in dB) is linearly dependent on the distance.
For the values given by the literature [2] for a non line of sight
(NLOS) environment, this model gives a greater path loss than
the other, so giving the worst SNR. The lower values of the
result make that some points of the graph are not shown. This is
due to the fact that the MAC layer does not accept such bad
SNR, discarding these packets.

The other models are hard to adjust because they summarize the
path loss in only one or two parameters which can not be
obtained by looking at the map. To adjust these parameters, the
optimal values for particular environments have been taken [1],
[2].

Figure 7: Indoor scenario with referenced building

Figure 6: SNR comparison for the outdoor models


4

depends on an individual building but also on the rest of the


buildings and objects in the nodes environment [3].

In this second project one scenario per model has been created in
which the propagation model for the scenario and the particular
parameters for this have been configured.
As we see in Figure 7 the network created for the indoor
environment consists of three terminal nodes and one server.
The nodes have been placed in different rooms to simulate a real
environment and to have different parameters values. Each node
only communicates with the server to evaluate the SNR in a
single path.
From all the scenarios created for indoor propagation models,
Cost Multi Wall model reflects better the differences between the
SNR of the three terminals due to the fact that parameters can be
better adjusted by using the information found in the referenced
map such as the number of penetrated walls or the type they
belong to.

Figure 9: Outdoor scenario with referenced building


The results of simulating this network are shown in Figure 9
where it can be observed that the SNR plots are very similar
when compared to the results of the indoor network shown
earlier. This similarity is due to the fact that the empirical
formula only takes into account the building percentage and the
free space loss.
Since these three nodes have the same building percentage
(10%) the only difference in the results is due to the distance
between each node and the server. This influences directly to the
free space propagation loss. The distances of the three nodes are
very similar and it makes the results very similar as it can be
seen in Figure 10.

Figure 8: SNR comparison between terminals


The results from the simulation of the Cost Multi-Wall model
reflect what was expected. The worst result from the three
stations is the SNR of the Terminal 1 due to the fact that it is
further away from the server and it has to penetrate more walls
than the others. It results in an average SNR of 28 dB.
Figure 7 also reflects that the SNR from Terminal 2 and 3 are
similar because they are closer and the average values of SNR
are greater than they are in Terminal 1 because the distance of
these two terminals from the server is shorter.
Another scenario has been created to simulate outdoor
environment using referenced maps. For outdoor models, the
referenced map is not important since these models do not have
parameters depending on the structure of the building. However,
the map is used to give our network a certain context.

Figure 10: SNR comparison between terminals

Conclusion and further work

For CCIR outdoor propagation model the only set up parameter


is the building percentage of the environment and it not only

In this paper we have presented new propagation models to


simulate wireless networks in indoor and outdoor environments.
5

These new indoor models allow us to simulate better wireless


networks. Furthermore, the goal of learning how to use all level
editors from OPNET Modeler to introduce the models has been
achieved. Firstly, simple models have been simulated to
understand their operation using different topology networks,
like Ad-hoc networks, IBSS and so on. Secondly, indoor and
outdoor models have been implemented to include them in our
new WLAN model.
After an analysis of previous papers and after seeing what
OPNET Modeler allow us to do we have chosen an
implementation based on empirical models. These algorithms
are compliant with short computation time and minimum
network overhead requirements. Contrary to this, as we have
seen the accuracy of the models is very low. Out of all the
models, we can say that cost multi-wall and motley-keenan
models are the most precise models.
The introduction of deterministic models in wireless nodes
remains an issue for further work. Thus, integrating data bases of
building with OPNET Modeler is a necessary issue to
accomplish it.
Following the steps of a preceding paper called Modelling La
Salle Universitys network using OPNET Modeler, also written
in this department, this work has continued improving the
network modeled there, specifically those parts where the access
technology is WLAN [5]. By doing so, we can achieve a more
accurate network model which will allow us detect more
problems in our network such as the coverage range of these
WLAN access networks.
References
[1] COST (European Co-operation in the Field of Scientific and
Technical Research), COST 231 Book, Final Report. Chapter 4,
Propagation Prediction Models.
[2] J.S Lee and L.E Miller, CDMA Systems Engineering Handbook.
Boston : Artech House, 1998
[3] Winprop Documentation: Software tool for the Planning of Radio
Communications Networks (Indoor),
http://www.awe-communications.com
[4] Soliman A. Al-Wabie, The New Wireless Local Area Networks
(WLANs) Standard. University of Maryland, 2002
[5] T. Fulgueira, Modeling La Salle Universitys network using
OPNET Modeler, La Salle University, 2003
[6] OPNET Modeler User Manual, Wireless Module Release 9.1
[7] OPNET Modeler on-line Documentation, Release 8.0

You might also like