Modeling and Simulation of near-earth Wireless
Sensor Networks for Agriculture based Application
L.M.Kamarudin, R.B.Ahmad, B.L.Ong, A.Zakaria, D.Ndzi
Embedded Computing Research Cluster,
School of Computer and Communication Engineering,
Universiti Malaysia Perlis,
Jalan Pangkalan Assam, 01000, Kangar, Perlis, Malaysia
Abstract—In recent years, there have been a number of
reported studies on the design of communication protocols
using simulation platform. However, most of the reported works
were evaluated using simple or idealistic wireless communica-
tion channel modeling. Experimental results have shown that
the characterization and modeling of wireless communication
channel is important to achieve a successful implementation of
wireless sensor network (WSN) systems in agricultural based
application. This paper investigates the impact of propagation
model towards WSNs system under OMNeT++ simulation en-
vironment. Several realistic propagation models for WSNs are
also reviewed. Several well known empirical vegetation models,
namely MED Weissberger Model and ITU-Recommendation
model are implemented in OMNeT++ simulation platform. It is
observed that propagation model used gives significant impact
towards the network performances. The results show that a
combination of plain earth (PE) and vegetation model give more
realistic result and can best describe the behavior of actual
WSN systems when deployed in a real environment. Antenna
heights and vegetation density are important parameters that
affect communication network coverage and connectivity.
Wireless sensor network (WSN) technology has a wide
range of potential applications including precision agriculture,
target tracking and emergency relief. Such networks consist
of large number of distributed nodes and those nodes are
expected to work independently in a very harsh environment.
Therefore, it is very difficult and time consuming to perform an
experiments with a large number of sensor nodes in different
kind of environments. Hence, the simulation platform is a
common way to evaluate and investigate the performance
of WSNs. Since most researchers focus on communication
protocols and applications, these parts of the simulation tools
are the most advanced and up-to-date. However, modeling of
the wireless communication channel is often too simplistic.
Detailed statistics about the publications at top conferences in
 prove that the papers with simple wireless communication
channel models outnumber others significantly. The risk is
that most communication protocols are designed based on
these simple models which are often too optimistic about the
performance of wireless communication systems in a wide
range of environments. An article in IEEE Communications
stated that “An opinion is spreading that one cannot rely
on the majority of the published results on performance
evaluation studies of telecommunication networks based on
stochastic simulation, since they lack credibility”. Thus,
the characterization and modeling of wireless communication
channel in simulation tools is important to achieve a successful
implementation of WSN systems.
In simulation environment, wireless communication system
performance is estimated based on radio propagation channel
models. Most of the models that are used in simulators
assume an obstacle free channel, and hence, a clear line of
sight between communicating nodes. The results produced
often poorly reflect the real scenario, such as plantation
area, in which the presence of vegetation significantly affects
communication between the nodes. As suggested in , for
short range near ground plantation environment, propagation
loss is modeled by an integration of the foliage imposed
effect and the effect from the radio wave reflections from the
ground and possibly tree canopy. There are various vegetation
propagation models reported in published literature , ,
 and these models are verified by numerous experimental
studies in , , . In this paper, two popular empirical
vegetation attenuation models, namely Weissberger Modified
Exponential Decay (MED) and ITU-Recommendation (ITU-
R) models, to predict propagation losses due to the existence
of trees within the communication path are implemented in
OMNeT++ simulation package.
This paper provides a review of propagation models suitable
for WSNs research, specifically in simulation studies. The
remainder of this paper is structured as follows:- Section III
presents the modelings of two vegetation propagation model
in OMNeT++ simulator. Section IV provides simulation setup
and discussed the findings regarding the effect of propagation
model on the network connectivity and communication range.
Finally, conclusions are provided in section V.
II. RADIO PROPAGATION MODEL IN SIMULATION TOOLS
In simulation environments, propagation losses due to wire-
less channel characteristics are calculated using propagation
models. These models are based on real measurements and
2010 International Conference on Computer Applications and Industrial Electronics (ICCAIE 2010), December 5-7, 2010, Kuala Lumpur, Malaysia
978-1-4244-9055-4/10/$26.00 ©2010 IEEE 131
represent a statistical mean or median of the expected path
loss. By predicting the average received signal power at a
given distance from a transmitting node, propagation models
estimate the communication coverage and calculate the success
or failure rate of message reception.
Most simulators assume a fix propagation distance for each
node where the signal propagates exactly d meters and no
further or a highly idealized propagation model such as the
Free Space Loss (FSL). Due to the presence of objects, radio
wave propagation is far more complicated than presented by
these simple models. Therefore more realistic propagation
models that can describe the property of targeted environments
are required to simulate WSN applications.
Based on the fact that each individual communication
system has to encounter different terrain, path, obstructions,
atmospheric conditions and other phenomena, it is impossible
to formulate the exact loss for all communication systems
in a single mathematical equation. As a result, different
models exist for different types of radio links under different
Basically, propagation models can be classified into deter-
ministic model or probabilistic model .
A. Deterministic Model:
In a deterministic model, the received signal power, Pr is
calculated based on actual properties of the environment such
as the distance, d between a transmitter and a receiver. A sim-
ple deterministic model only considers the distance between
nodes as a parameter models whereas a more complex model
takes into account multipath fading effects in the environment
as a parameter to be calculated.
1) Free Space Loss (FSL) Propagation Model: FSL is a
widely used model in simulation environments that assumes
an ideal propagation condition. FSL is suitable for predicting
the signal strength at the receiving node when there is a
clear Line of Sight (LOS) path between the transmitting and
receiving nodes . The received signal power decreases
with increasing distance between the transceivers and is given
by the Friis Equation  which is represented by (1).
Pr=Pt× Gt× Gr× λ2
where Ptis the transmitted signal power, Pris the received
signal power, Gt and Gr are the transmitter and receiver
antenna gains respectively and λ is the wavelength. Path loss
can be calculated by (2).
LFSL= −27.56 + 20log(d) + 20log(f)
where f is the frequency in MHz and d is the distance
between the isotropic antennas in meters.
However, for obstructed paths, it is not adequate to simply
use FSL model to predict the signal strength when radio waves
propagate near the ground.
2) Plane Earth (PE) Propagation Model: Typically, sensor
nodes are placed near ground area (0~3 meters above ground)
with low transceiver antenna heights . Therefore, when
radio waves propagate near the ground, the path loss can be
better described by the plane earth (PE) model  rather than
the FSL model.
The PE model assumes that the received signal strength
is the sum of the direct line of sight propagation path and
one ground reflected component between the source and the
destination nodes as shown in Fig. 1. The received signal
strength at distance, d is predicted by (3) , .
Pr=Pt× Gt× Gr× h2
Where ht and hr are the transmitter and receiver antenna
heights, respectively. Path loss can be calculated by (4).
LPE(dB) = 40log10(d) − 20log10(ht) − 20log10(hr) (4)
As a plane terrain appears, ground reflection may occur.
Therefore, to maximize the connectivity, the antenna height
must clear the 1st Fresnel zone. The strongest signal compo-
nent is the direct line between transceivers and always lies in
the 1st Fresnel Zone . The Fresnel zone radius, r can be
calculated by (5).
r = 17.32
where, d is the total distance between antennas in km and
f is the frequency in GHz.
3) Vegetation Propagation Model: For most applications in
agriculture, a typical property of the propagation environment
is the presence of vegetation that acts as obstacles in the radio
path causing multiple scattering, diffraction, and absorption.
These can result in severe fading of the received signal
strength, and produce an excess vegetation induced loss. These
losses have to be considered in the wireless communication
channel modeling .
ITU  stated that, existence of vegetation in the middle
of propagation path attenuates radio waves considerably espe-
cially at higher frequencies. Attenuation caused by vegetation
is approximately 0.4dB/m at 3GHz, 0.1dB/m at 1GHz, and
0.05dB/m at 200MHz. Since WSNs usually operated at
2.4GHz, the effect of vegetation could be considered to be
Many studies have been carried out to characterize and
model the effects of vegetation experimentally. They have been
reviewed and summarized into several well-known empirical
vegetation propagation models for the near ground propa-
gation, such as Weissberger’s Modified Exponential Decay
model  and ITU Recommendation (ITU-R) . Attenuation
predicted by the vegetation propagation model is in addition to
other type of non foliage loss such as direct wave and reflected
• Weissberger’s Modified Exponential Decay Model :
This model is suitable to be use when the propagation
path is blocked by dense, dry, trees with leaves within
400 m depth. This model is applicable in situations
where propagation is likely to occur through a body
of trees rather than by diffraction over the top of the
trees. Thus, this model is considered to be appropriate to
simulate WSNs applications in agriculture environment
since nodes are usually placed near the ground. Radio
frequency ranges suitable for with this model is between
230MHz to 95GHz. The formula is given by (6).
14 < df≤ 400
0 < df≤ 14
where L is the losses due to foliage, f is the transmission
frequency in gigahertz (GHz) and df is vegetation depth
between transmitter and receiver in meters (m).
• ITU Recommendation (ITU-R):
ITU-R model was developed from measurements carried
out mainly at Ultra High Frequency (UHF), and was
proposed for cases where either the transmitter or the
receiver is near to a small (d < 400m) grove of trees
such that the majority of the signal propagates through
the trees. The model is represented by (7).
LITU−R(dB) = 0.2f0.3d0.6
where f is the transmission frequency in megahertz (MHz)
• ITU Terrestrial Model with One Terminal in Woodland
This model is suitable to be used when either one of
the sensor nodes is inside the vegetation. The excess
attenuation, Aev due to the presence of the vegetation
is given by (8).
Aev= Am[1 − exp(−dfγ/Am)]
where df is the vegetation depth within woodland in
meters (m),γ is the initial vegetation attenuation curve
gradient at small vegetative depth (dB/m), Am is the
maximum attenuation for one terminal within a specific
type and depth of vegetation in decibels (dB).
B. Probabilistic Model
In most environments, propagation loss is dynamic and can
be characterized statistically. Therefore, probabilistic models
allow a more realistic modeling of radio wave propagation.
A probabilistic model takes a suitable deterministic model
based on local environment such as MED model in vegetation
environments or FSL model as one of its input parameters
in order to get a mean at each propagation range. For every
node, the received power is then drawn from a distribution.
The results give a more diverse distribution of successful
receptions. It is possible that on the balance of probability; two
nodes that are close to each other cannot communicate while
a pair of nodes far apart over the deterministic communication
Table I: Values for different environment
Urban area, LOS
Urban area, non-LOS
(a) Path loss exponent α
2.7 - 5
1.6 - 1.8
4 - 6
Office, hard partition
Office, soft partition
(b) Shadowing variance σ
4 - 12
3 - 6
range can communicate. The distribution of these effects
depend on the probabilistic model and its parameters.
1) Log-Normal Shadowing: : Shadowing occurs when ob-
jects blocked LOS between transmitter and receiver. Therefore,
a simple statistical model can account for unpredictable “shad-
owing”. This method assumes that the average received signal
strength decreases logarithmically with distance. This model
can be used in simulation to estimate received signal strength
for random locations in the system. This model uses a normal
distribution with variance σ as in Table IIb to distribute receive
power in the logarithmic domain. The received signal strength
is given by (9):
Pr= Prdet(d0) × 10LS(d)
where Prdetis a deterministic model at reference distance
dosuch in (1) or (6) and LS(d) is the path loss factor in dB
given by (10).
2) Rayleigh distribution: : Rayleigh is a fading model that
describe the time-correlation of the received signal strength.
This propagation model represent situation with non Line
of Sight (NLOS) and only scattered components exist. This
model shows intensive variations in received signal strength
because the scattered signal components can either combine
constructively or destructively. Similar to Log-Normal shad-
owing, this model greatly depends on the environment and the
type of network. Therefore, a suitable deterministic model is
used with this model. The received signal power is defined by
Pr(d) = Prdet(d0) × 10PL(d)× log(1 − unif(0,1))
where the path loss is given by (12).
Lrayleigh(d)[dB] = −αlog10
The value of α is based on Table IIa.
III. MODELING OF VEGETATION PROPAGATION MODEL IN
In this section, the vegetation propagation models developed
in OMNeT++ simulation environment is briefly discussed.
Two vegetation models namely MED and ITU-R models are
modeled and tested.
A. OMNeT++ Simulation Platform
The Objective Module Network Test-bed in C++ (OM-
NeT++)  is an open-source simulation environment devel-
oped at the Technical University of Budapest by Andras Varga.
It provides a component-based and discrete event simulation
environment, which can be combined and reused flexibly.
Additionally, OMNeT++ provides strong GUI support and an
embedded simulation kernel. The primary application area of
OMNeT++ is the simulation of computer networks. It has been
shown that this simulation framework is suitable for simu-
lating WSNs owing to its modular structure and using NED
language for configuration. OMNeT++ offers various open-
source communication network simulation libraries, namely
INET and Mobility Framework. Both frameworks are suitable
to simulate wired, wireless and adhoc networks.
INET Framework only offers the FSL Propagation model
for signal strength determination. Mobility Framework has
been extended in  to support probabilistic propagation
models namely Rayleigh, Nakagami and Rice models. Yet,
both frameworks do not support any propagation model that
calculates extra attenuation due to the presence of obstacles
such as vegetation in the propagation path.
The modeling of propagation through vegetation commu-
nication channel in INET Framework are through two mod-
ules. The modules are Channel Control and IEEE802.15.4Phy
which are located in Network Interface Card (NIC) module
of each node. The Channel Control module maintains all
possible connections between nodes in the network. Nodes
are connected if d is less than maximum propagation distance,
dmax. The existence of a radio link between nodes is computed
by comparing the Pr calculated using propagation model to
the Receiver Sensitivity threshold (Rs).
The IEEE802.15.4Phy module role in each node is to
calculate Prfor every incoming message and update the signal
to interference and noise ratio (SNIR). For this, a new sub-
class (PLPropagationReceptionModel) that is used to calculate
the Prhas been created. In case of the existence of vegetation
in the link path, the overall path loss, (PLa) is the sum of
plane earth (PE) path loss in (4) and the vegetation loss as in
(6) or (7). Therefore, Pris computed by subtracting Ptfrom
Based on Pr, the potential signal reception is distinguished
between two different cases:
1) The message is correctly received by a receiving node if
Pris greater than Rsand no other messages are being
sent or received.
2) In case Pris less than Rs, messages are treated as noise.
Figure 1: Near earth propagation for WSNs over foliage
An integration into the Mobility Framework is straight for-
ward, since the INET Framework is based on the Mobility
Framework. It shares the Channel Control class so that the
integration described above applies in the same way. A notable
difference is that the IEEE802.15.4Phy module in Mobility
Framework is known as SnrEval. The SnrEval Module role is
similar to IEEE802.15.4Phy module role in INET Framework
that predicts the signal strength of incoming message. Hence,
the functions implemented are the same.
A. Simulation Setup
The environment investigated herein is for agriculture based
application as shown in Fig. 1, where a specified number of
sensors are deterministically distributed over 100x100m area.
The simulation consists of 100 nodes and the distance between
two adjacent nodes is 10m. The arrangement of wireless
nodes position in this simulation model is based on the actual
separation of palm trees in an oil palm plantation. The terrain
is nearly flat and consists mainly of soil and sand, with some
part covered by grass.
One of the main design issues of WSNs is the sensor
node placement. Sensors may be deployed either randomly or
deterministically depending on the application. However, node
placement without proper planning may result in no network
connectivity and severe network degradation. Poor positioning
could result in severe shadowing by tree trunks, a situation that
is highly like in vegetations that have large tree trunks such
as oil palms. A successful communication between nodes can
only be achieve if the nodes are within the communication
range. Therefore, a grid based deployment is best suited
agriculture based applications since the applications require a
continuous and precise monitoring. Equilateral triangle, square
grid and regular hexagon are popular regular pattern in grid
node deployment, and it is proved to be most effective when
the base station (BS) is located in the centre of a grid.
Thus, this simulation study employ a stationary square grid
topology with the BS located is in the middle of the topology.
The IEEE 802.15.4 standard developed in  is used with
operating frequency of 2.4GHz. The experiments run in a
beacon enabled star environment. The wireless channel bit
rate is set to 250kbps. Simulations are run at fixed duration
Figure 2: Connectivity for different Propagation Model at
various foliage depth
of 2000s with packet generated at constant bit rate of 30s. All
nodes are configured to have a fixed transmission power of
1mW and a sensitivity value of -85dBm based on Chipcon
CC2420 transceiver chip. Omnidirectional antennas with a
typical gain of 1 dBi are used at in the transmitter and receiver.
The outcome of this simulation is to investigate the effect of
varying vegetation depth and antenna heights on communica-
tion network coverage and the connectivity of the nodes using
different propagation models. Five types of radio propagation
model are simulated. The existing FSL model in OMNeT++
simulation tool is compared with other propagation models
that consider vegetation density. These include MED, ITU-R,
and a combination of vegetation attenuation models with plane
earth model (PE+MED /PE+ITU-R). Simulation studies also
cover the effect of antenna heights.
B. Result and Discussion
This section discuss different propagation model impact on
varying setup of WSNs, i.e node placement and communi-
cation network coverage to ensure the connectivity between
the sensor nodes and the BS. The results are compared with
widely used models in simulation platform for example, the
1) Node Placement and Network Connectivity: The star
topology of WSNs requires all nodes to be able to send
data directly to the BS. The network connectivity simulation
identified the sensor nodes that are within range of the
BS. Therefore, the percentage C of the network connectivity
between nodes and the BS is as follows:
where Nc is the connected nodes with the BS and N is
number of nodes in the topology.
The simulation is performed using five different propagation
models at three different antenna heights and various vegeta-
tion depths. The simulation results are shown in Fig. 2.
The result shows that FSL gives 100% of connectivity
despite the presence of vegetation. This is not surprising, given
Figure 3: Communication Coverage for difference Propagation
model at various foliage Depth
that this model is an idealistic model that does not consider
the existence of obstacles and antenna heights which can affect
the propagation behaviors. MED and ITU-R models show a
decreasing connectivity with an increasing vegetation depth.
However, no changes can be observed with varying antenna
height. Based on experimental studies, it has been reported in
 that propagation model for near ground propagation is the
sum of the foliage effect and the ground reflection effect. The
PE+MED model at low antenna height of 0.1m show a very
low connectivity in which BS can only cover approximately
4 out of 100 nodes. This is true for most of the cases in
WSNs where nodes are placed on the ground. As the antenna
height increase, connectivity improves. At 1m antenna height,
PE+MED model shows 95% connectivity when the foliage
depth is 10m and a perfect connectivity is achieved when
foliage depth is less than 6m. It can be observed that PE+ITU-
R model gives lower connectivity compares to PE+MED
model at all antenna height. The analytical calculation based
on equation 5 proves that antenna height must be above 0.5m
to clear the ground reflection. However, the existence of grass
has not been considered in the calculation. Thus, based on
the results discussed above, the antenna of each sensor node
should be placed higher above the ground in order to ensure
optimum connectivity between the nodes.
2) Communication Coverage: Typically, sensor nodes are
deployed based on the maximum communication range. There-
fore, the larger the communication range of a node, the further
it can be deploy. From datasheets, the maximum communica-
tion range of TelosB motes is between 75 m and 100 m, and
IRIS is about 500 m  if there are no obstacles between
the communicating nodes. However, in many cases where
they have been used, communication failed because proper
studies were not carried out, to estimate the communication
coverage of the node before deployment in the real targeted
Various vegetation depths between BS and sensor node have
been used to investigate the the network performance and
the communication coverage. Fig. 3 displays the simulated
result. The graph for FSL shows that communication coverage
is uniform at 180m despite of increasing vegetation depth.
6 Download full-text
On the other hand, communication range for MED and ITU-
R models are decreasing with increasing vegetation depth.
MED+PE model at 1m antenna height shows a decreasing
communication range from 138m to 98m at vegetation depths
of 0 and 10m, respectively. PE+MED and PE+ITU-R models
at an antenna height of 0.1 m shows that communication range
is below 20m at all vegetation depth. This proves that the
height of the antenna significantly affect communication range
and thus, will affect network performances.
In this paper, several published works regarding propagation
model for WSNs system have been reviewed. Two empirical
vegetation propagation models, namely MED Weissberger and
ITU-R model have been developed in OMNeT++ simulation
package. The effect of foliage on WSN systems’ performances
is studied and discussed.
From the result, the widely used FSL model shows over
optimistic result despite the existence of vegetation in the
transmission path. The real radio communications channel
is much more complex than that and different propagation
model give different results. Even at close range, result shows
that the path loss can be considerable as the radio wave is
attenuated by in vegetation and reflected by the ground. Thus,
it is essential to use an appropriate propagation model based
on the environmental condition since the evaluation of new
protocols using an inappropriate model may lead to wrong
Result shows that antenna height is one of the most
important considerations when implementing a WSN system.
Short range WSN system with antennas placed at or near
the earth’s surface usually fail when implemented in the real
environment. This type of sensor network needs a proper
node placement strategy to maintain network connectivity
between sensor nodes. Therefore, adequate propagation model
should be used to predict antenna heights and path loss to
ensure that the system maintain link connectivity.
 D. Kotz, C. Newport, R. S. Gray, J. Liu, Y. Yuan, and C. Elliott, “Ex-
perimental evaluation of wireless simulation assumptions,” in MSWiM
’04: Proceedings of the 7th ACM international symposium on Modeling,
analysis and simulation of wireless and mobile systems, (New York, NY,
USA), pp. 78–82, ACM Press, 2004.
 K. Pawlikowski, H. d. Joshua Jeong, and J. s. Ruth Lee, “On credibility
of simulation studies of telecommunication networks,” IEEE Communi-
cations Magazine, vol. 40, pp. 132–139, 2002.
 Y.S.Meng, Y.H.Lee, and B.C.Ng, “Path loss modeling for near-ground
vhf radio-wave propagation through forests with tree-canopy reflection
effect,” Progress In Electromagnetics Research M, vol. 12, pp. 131–141,
 M. A. Weissberger, “An Initial Critical Summary of Models for Predict-
ing the Attenuation of Radio Waves by Trees,” tech. rep., July 1982.
 CCIR, “Influences of terrain irregularities and vegetation on tropospheric
propagation,” tech. rep., CCIR Report, July 1986.
 I.-R. R. P.833-2, “Attenuation in vegetation,” tech. rep., Int. Telecom
Union, Geneva, 1999.
 N. Savage, D. Ndzi, A. Seville, E. Vilar, and J. Austin, “Radio wave
propagation through vegetation: Factors influencing signal attenuation,”
Radio Sci., vol. 38, no. 5, p. 108, 2003.
 N. Moraitis, V. F. Milas, and P. Constantinou, “On the empirical model
comparison for the land mobile satellite channel,” in VTC Spring,
pp. 1405–1409, 2007.
 A. Kuntz, F. Schmidt-eisenlohr, O. Graute, H. Hartenstein, and M. Zit-
terbart, “Introducing probabilistic radio propagation models in omnet++
mobility framework and cross validation check with ns-2,” 2008.
 J. S. Seybold, Introduction to RF Propagation. Wiley-IEEE, 2005.
 T. S. Rappaport, Wireless Communications: Principles and Practice (2nd
Edition). Prentice Hall PTR, 2 ed., January 2002.
 W. C. Lee, Mobile Communications Engineering: Theory and Applica-
tions. McGraw-Hill Professional, 2nd ed., 1998.
 A. Varga, “The omnet++ discrete event simulation system,” in Pro-
ceedings of the European Simulation Multiconference, (Prague, Czech
Republic), pp. 319–324, SCS – European Publishing House, June 2001.
 H. Liu, Z. Meng, and Y. Shang, “Sensor nodes placement for farmland
environmental monitoring applications,” in WiCOM’09: Proceedings of
the 5th International Conference on Wireless communications, network-
ing and mobile computing, (Piscataway, NJ, USA), pp. 3193–3196, IEEE
 F. Chen and F. Dressler, “A Simulation Model of IEEE 802.15.4 in
OMNeT++,” in 6. GI/ITG KuVS Fachgespräch Drahtlose Sensornetze,
Poster Session, (Aachen, Germany), pp. 35–38, July 2007.
 Z. Alliance, IRIS Wireless Measurement System. MEMSIC Inc., 6020-
0124-02 rev a ed.