Sei sulla pagina 1di 7

www.ijape.

org InternationalJournalofAutomationandPowerEngineering(IJAPE)Volume2Issue4,May2013
116
AChannelModelforVANETSimulation
System
LimingZhang
1
,FengChen
*2

DepartmentofAutomation,UniversityofScienceandTechnologyofChina
Hefei,Anhui,China
1
zlm0913@mail.ustc.edu.cn;
*2
chenfeng@ustc.edu.cn

Abstract
Vehicular Ad Hoc Network (VANET) is the key technology
of Vehicle Infrastructure Integration (VII). Microscopic
simulation is an effective means for VANET research, and
radio channel model is the basis of VANET simulation. In
this paper, radio wave propagation models, such as
reflectionanddiffractionmodels,arecombinedtodevelopa
VANET channel model, and the model is introduced into
VANET simulation system. Path loss and received power
under different traffic flow are analysed using our VANET
simulationsystem.Thesimulationresultsareconsistentwith
the real situations of VehicletoVehicle (V2V)
communication.
Keywords
VehicularAdHocNetwork;Simulation;ChannelModel;PathLoss;
ReceivedPower
I nt r oduc t i on
VANET is the development trend of Intelligent
Transportation System (ITS). It can significantly
improve the traffic in safety, efficiency and
informationservice.VANEThasreceivedconsiderable
attention in recent years. The standard of Dedicated
Short Range Communications (DSRC) has been
established.FederalCommunicationsCommissionhas
extended IEEE 802.11 Standard for VANET. This
extended standard is called IEEE 802.11p, which is
authorized to work in 5.9GHz free band. It is highly
difficult to study VANET by field experiments due to
its complexity. Traffic simulation can describe
temporal and space behaviours of vehicles in
microlevel with low cost, so it becomes an effective
meansfortheresearchofVANET.
Channel model is the basis of VANET simulation.
VANET channel is affected by multiple factors
including the buildings, road situations, vehicle type,
and vehicle relative velocity. The related works are as
follows: A VANET channel is modelled using optical
ray tracking technique by Maurer. This kind of model
is usually in good agreement with the real world, but
implemented with high computational complexity. A
widely used simplified stochastic model is employed
to fit to realworld measurements by Turkka and
Paschalidis.Thiskindofmodeloffersapproximationof
VANET channel at low computational cost, but is not
suitable for typical scenarios. In order to simulate the
VANETchannelintypicalscenarios,manyresearchers
consider buildings as obstacles, especially at the
intersection. However, the impact of vehicles as
obstacles on V2V communication has been largely
neglected in VANET research. Boban emphasized the
importance of considering other vehicles on the road
as obstacles, and presented a method to analyse the
existence of the LineofSight (LOS) and
NonLineofSight(NLOS).
In this paper, the vehicles between transmitter and
receiver are considered as obstacles. By combining
severalradiowavepropagationmodels,weproposea
VANET channel model which is easy to implement
andsuitableforspecifictrafficscenarios,andapplythe
modelintoVANETsimulationsystem.
The paper is organized as follows. In Section 2, path
loss and radio wave propagation models are
introduced.InSection3,theVANETchannelmodelis
developed,andthesimulationprocedureisdescribed.
In Section 4, the simulation results are analysed.
Finally,theconclusionsaredrawninSection5.
Rel at ed Theor i es
Radio waves may encounter many obstacles while
propagatingfromtransmittertoreceiver.Thiswilllead
to reflection, diffraction and scattering of radio wave.
Both largescale propagation effect and smallscale
propagation effect will occur when radio waves
propagate. Largescale propagation effect is used to
describe the variations of field strength received over
relatively long distances, in which path loss and
shadowing are mainly considered. Smallscale
InternationalJournalofAutomationandPowerEngineering(IJAPE)Volume2Issue4,May2013 www.ijape.org
117
propagationeffectisusedtodescribethevariationsof
field strength received over very short distances, in
which multipath delay and Doppler shift are mainly
considered.
Theconceptofpathlossplaysavitalroleindescribing
theattenuationofsignalpower.InVANETsimulation,
to calculate the received power, we consider the path
lossinseveralradiowavepropagationmodels.
PathLoss
Pathlossisdefinedas
=
10
[ ] 10log
t
r
P
PL dB
P
(1)
where,
t
P isthetransmittedpower;
r
P isthereceived
power.
According to the theory and measurements, received
power attenuates with the logarithm of propagation
distance, which can be expressed as LogDistance
PowerLaw(LDPL):
= +
0
0
[ ] ( ) 10 log( )
d
PL dB PL d n
d
(2)
where, n is the path loss exponent, to describe how
quicklythesignalpowerattenuates; d isthedistance
between Transmitter and Receiver (TR);
0
d is the
reference distance determined by measurements. The
valueof n dependsonthepropagationenvironment.
Infreespace, = 2 n ,otherwise, > 2 n .
The statistical model can be used when there are
obstacles with unknown sizes and locations between
TR. The most widely used statistical model is
LogDistanceShadowingModel:
o
= + +
0
0
[ ] ( ) 10 log( )
d
PL dB PL d n X
d
(3)
where,
o
X is a normally distributed random variable
withmean0andstandarddeviation o ,describingthe
shadowing effect caused by obstacles. The value of n
and o can be obtained by linear regression of the
measurementdata.
RadioWavePropagationModels
The following are several radio wave propagation
models including free space model, tworay ground
model, single knifeedge diffraction model and
multipleknifeedgesdiffractionmodel.Weaddressthe
pathlossforallofthesemodels.
1) FreeSpaceModel
Freespacemodelisusedwhenthereisnoobstacle
betweenTR,andtheradiowavepropagatesalong
astraightline.
Thereceivedpoweratdistance d fromtransmitter
is

t
=
2
2 2
(4 )
l
r t
G
P P
d
(4)
where,
t
P is the transmitted power;
r
P is the
received power; =
1 2 l
G G G is the product of
transmitting antenna gain and receiving antenna
gain; d istheTRdistance; isthewavelength.
Thepathlossoffreespacemodelis

t
=
10
[ ] 20log
4
l
G
PL dB
d
(5)
2) TwoRayGroundModel
The received signal in tworay ground model
consistsoftwocomponents:thesignalpropagating
through free space and the signal reflected off the
ground.
Thereceivedpoweratdistance d fromtransmitter
is
|

t
A
= +
+
2
2

( )
4
j
l r
r t
G G e
P P R
l x x
(6)
where,
l
G and
r
G are the products of antenna
gains;
l isthelengthofdirectpath;
+

x x isthelengthofreflectionpath;
| A is the phase difference between the two
receivedsignalcomponents(denotedbyEq.7)
t
|

+
A =

2 ( ) x x l
(7)
R is the ground reflection coefficient. There are
two cases in reflection including vertical
polarizationandhorizontalpolarization.Inthecase
ofverticalpolarization,
u c u
u c u

=
+
2
2
sin cos
sin cos
r
r
R
(8)
with c
r
isthedielectricconstantoftheground.
Thepathlossoftworaygroundmodelis
www.ijape.org InternationalJournalofAutomationandPowerEngineering(IJAPE)Volume2Issue4,May2013
118
|

t
A
= +
+
10
[ ] 20log
4
j
l r
G G e
PL dB R
l x x
(9)
3) SingleKnifeEdgeDiffractionModel
WhenthereisanobstaclebetweenTR,theshapeof
the obstacle can be viewed as knifeedge for
calculatingthelosscausedbydiffraction.
TheFresnelKirchoffdiffractionparameter v is

= +
1 2
2 1 1
( ) v h
d d
(10)
where, h is the height of the top of the obstacle
above the straight line joining the two ends of the
path,iftheheightisbelowthisline, h isnegative.
1
d and
2
d arethedistancesofthetwoendsofthe
path from the top of the obstacle. is the
wavelength.
When the first Fresnel zone is unobstructed
( s 0.78 v ), the effect caused by diffraction can be
neglected.Theapproximationofthelosscausedby
diffractioncanbeexpressedasafunctionof v :

+ + + >
=

s

2
10
6.9 20log ( ( 0.1) 1 0.1) 0.78
( ) [ ]
0 0.78
v v v
J v dB
v
(11)
4) MultipleKnifeEdgesDiffractionModel
There may be several separate obstacles between
TR.Iftherearemorethanthreeobstacles,theeffect
causedbyeachobstaclewillbeevaluated,andonly
the larger three will be employed to calculate the
loss.
Assuming the number of obstacles is n , the
FresnelKirchoff diffraction parameter
i
v of
obstacle i is

= +
1 2
2 1 1
( )
i i
i i
v h
d d
(12)
Theobstaclewiththehighestvalueof v istermed
astheprincipaledge p ,andthecorrespondingloss
is ( )
p
J v .If s 0.78
p
v ,thetotaldiffractionlossis0;
otherwise > 0.78
p
v ,theprocedureisdescribedas
follows:
Obtain the largest
t
v with loss ( )
t
J v from the
transmittertopoint p ;
Obtainthelargest
r
v withloss ( )
r
J v frompoint
p tothereceiver;
Calculatethetotaldiffractionloss.
Thetotaldiffractionlossis
+ + + >

=

s

( ) [ ( ) ( ) ] 0.78
[ ]
0 0.78
p t r p
p
J v T J v J v C v
J dB
v
(13)
where,
= 1.0 exp[ ( ) / 6.0]
p
T J v
(14)
= + 10.0 0.04 C D (15)
D isthetotalpathlength(km).
Vanet Channel Model i ng
VANETChannelAnalysis
In VANET, the communication between vehicles may
be interfered by other vehicles as the obstacles. If the
communication is not interfered, the signal will
propagateinLOSpath,otherwiseinNLOSpath.Inthe
condition of NLOS, the signal power attenuates
significantly.
The antenna is usually installed on the roof of the
vehicle. The LOS may exist even if there are other
vehiclesbetweenTR.AsshowninFig.1,theheightof
the obstacle is below the straight line joining the TR
antennas.Asageneralrule,iftherstFresnelellipsoid
is not obstructed, the effect of the diffraction can be
neglected,anditcanberegardedasLOS.

FIG.1THELOSEXISTSWHENTHEREISANOBSTACLE
BETWEENTR
In VANET simulation, the channel model should be
established in the condition of LOS and NLOS
respectively. On the one hand, in the real traffic, the
V2V communication is switching between LOS and
NLOS constantly. Its not suitable to describe the
propagation of signal and the effect of obstacles using
thestatisticalmodel.Ontheotherhand,inthespecific
scenario,thesizesandlocationsofvehiclesorobstacles
areavailable,itcanbedeterminedwhetherthereexists
LOS or not. When calculating the path loss, different
modelsshouldbeadoptedaccordingtowhetherthere
existsLOSornot.
When there is no vehicle between TR, the direct
component and the ground reflection component are
InternationalJournalofAutomationandPowerEngineering(IJAPE)Volume2Issue4,May2013 www.ijape.org
119
mainly considered. When there are one or more
vehicles between TR, it needs to determine whether
there exists LOS. If LOS exists, only the direct
componentisconsidered,otherwisethediffractionloss
caused by obstacles need to be considered. The effect
of reflection and scattering from the surrounding
objects are relatively weak, so not included in this
paper.
LOS/NLOSJudgingandPathLossCalculation
In specific scenarios, the information of a vehicle can
be obtained at any time, including the road belonged,
thecoordinateofthevehicle,thesizeofthevehicleand
the height of the antenna. The information is used to
judgeLOS/NLOSandcalculatethepathloss.
( , )
i i
x y


FIG.2LOS/NLOSJUDGINGWHENTHEREAREOBSTACLES
BETWEENTR
Thecoordinateofvehicle i isdonatedas ( , )
i i
x y ,the
mean value of its size is donated as
i
l , the height of
theantennaisdonatedas
i
h ,thedistancefromvehicle
i tothestraightlinejoiningTRisdonatedas
i
d ,the
middlepointofthestraightlinejoiningTRisdonated
as
0 0
( , ) x y , the distance between transmitter and
middle point is donated as r , as shown in Fig.2. The
procedureofLOS/NLOSjudgingisasfollows:
Step 1. Judge whether the straight line joining TR is
obstructed by other vehicles or not. The method is:
Makeacirclewithcenterat
0 0
( , ) x y andradiusequal
to r . In the simulation scenario, find all the vehicles
meet + <
2 2
0 0
( ) ( )
i i
x x y y r and <
i i
d l . If no
vehiclemeetsthecondition,gotoStep6;otherwisego
tothenextstep.
Step 2. For the vehicles obstructing the straight line
joining TR, calculate the FresnelKirchoff diffraction
parameter
i
v according to
i
h . When the rst Fresnel
ellipsoid is unobstructed ( s 0.78
i
v ), go to Step 5;
when the rst Fresnel ellipsoid is obstructed by only
one vehicle( > 0.78
i
v ) , go to Step 4; otherwisego to
thenextstep.
Step 3. The situation is considered as NLOS. Use
multiple knifeedgesdiffraction model to calculate the
diffraction loss, and add the result to the path loss
calculatedbyfreespacemodel.
Step4.ThesituationisconsideredasNLOS.Usesingle
knifeedge diffraction model to calculate the
diffraction loss, and add the result to the path loss
calculatedbyfreespacemodel.
Step 5. The situation is considered as LOS. Use free
spacemodeltocalculatethepathloss.
Step6.ThesituationisconsideredasLOS.Usetworay
groundmodeltocalculatethepathloss.
SimulationProcedure
For the specific transmitter and receiver (two cars in
simulationscenario),theprocedureofcalculatingpath
lossandreceivedpowerisshownasFig.3:
Update the coordinates of
vehicles according to
vehicle behavior models
Simulation is
terminated
Yes
No
Output
Get the information of
vehicles around the
transmitter and receiver
Judge LOS/NLOS and
calculate path loss
Calculate the received
power by transmitted power
and path loss
End
Start

FIG.3THESIMULATIONPROCESSDIAGRAM
The received power can be acquired by transmitted
powerandpathloss:
www.ijape.org InternationalJournalofAutomationandPowerEngineering(IJAPE)Volume2Issue4,May2013
120
= [ ] [ ] [ ]
r t
P dBm P dBm PL dB
(16)
The received power threshold is employed to judge
whether accepting the packet from transmitter. If the
receivedpowerislargerthanthethreshold,thepacket
isaccepted;otherwisethepacketisdiscarded.
Si mul at i on Resul t s and Di sc ussi on
SimulationScenariosandParameterSettings
In this paper, we implement and evaluate the
proposed channel model in USTCMTS2.0 platform.
USTCMTS2.0isaServiceOrientedArchitecture(SOA)
based Microscopic Traffic Simulation (MTS) system,
which was developed by DF & ITS Laboratory in
USTC. This platform has been utilized to evaluate
microscopic traffic models, and distributed dynamic
loadbalancingalgorithms.

FIG.4THESIMULATIONSCENARIOSWITHLIGHTERAND
HEAVIERTRAFFICFLOW
TABLE1THEPARAMETERSETTINGSINTHESIMULATION
Parameter Value
Trafficflow 500pcu/h,3000pcu/h
Vehiclelength 4m
Vehiclewidth 2m
Lanenumber 5
Lanewidth 3.5m
Generalvehicleantennaheight 1.41.8m
Transmitterantennaheight
t
h
1.5m
Receiverantennaheight
r
h
1.7m
Transmitterantennagain
1
G
2
Receiverantennagain
2
G
2
Transmitterpower

x
T Power
20dBm
Receivedpowerthreshold

x
R Threshold

95dB
Wavelength 0.051m
Dielectricconstantoftheground
r
e
1.02
Inordertoanalysethecommunicationperformanceof
VANET, we perform the simulation in two kinds of
typical scenarios with lighter and heavier traffic flow
respectively, as shown in Fig.4. The transmitter is
marked green, the receiver is marked yellow and
general vehicles are marked red. The parameter
settingsinsimulationscenariosarelistedinTable1.
Forthe5.9GHZcarrierfrequencyofIEEE802.11p,the
wavelength is 0.051 m. The dielectric constant of the
ground is set to 1.02. The received power threshold is
setto95dB.
SwitchingbetweenLOSandNLOS
Fig.5 shows the variation of received power in the
simulation of a vehicle passing through the straight
line joining TR. The received power is significantly
reducedunderNLOS.

FIG.5THEVARIATIONOFRECEIVEDPOWER
PathLoss
To evaluate the performance of proposed channel
model, we collect the path loss in different TR
distance as sample points in the simulation, and then
compare thestatistical characteristics of sample points
withrealworldmeasurementsusingtheLDPLmodel.
For this purpose, the sample points are analysed by
linearregressioninlogarithmiccoordinatesystem.
In the scenario with light traffic flow (500pcu/h), the
distribution of sample points is shown in Fig.6. The
linearregressionresultis = 2.013 n and o = 7.978 .

FIG.6THEDISTRIBUTIONOFSAMPLEPOINTSINLIGHT
TRAFFICFLOWSCENARIO
InternationalJournalofAutomationandPowerEngineering(IJAPE)Volume2Issue4,May2013 www.ijape.org
121
Inthescenariowithheavytrafficflow(3000pcu/h),the
distribution of sample points is shown in Fig.7. The
linearregressionresultis = 2.689 n ando = 6.205 .

FIG.7THEDISTRIBUTIONOFSAMPLEPOINTSINHEAVY
TRAFFICFLOWSCENARIO
In several literatures, n and o in LDPL are
calculatedbasedontherealworldmeasurements,and
the value of n is between1.5 ~ 5 , the value of o is
between 2 ~ 8 . The simulation result is consistent with
thesituationinreality.
ReceivedPower
In the scenario with light traffic flow (500pcu/h), the
distributionofreceivedpowerisshowninFig.8.

FIG.8THEDISTRIBUTIONOFRECEIVEDPOWERINLIGHT
TRAFFICFLOWSCENARIO

FIG.9THEDISTRIBUTIONOFRECEIVEDPOWERINHEAVY
TRAFFICFLOWSCENARIO
Inthescenariowithheavytrafficflow(3000pcu/h),the
distributionofreceivedpowerisshowninFig.9.
The simulation result shows, reliable communication
distancedecreasesalongwiththetrafficflowincrease.
Conc l usi on
In this paper, based on several typical radio wave
propagation models, we developed the VANET
channel model in the condition of LOS and NLOS
respectively. The model proposed is more suitable
than statistical model for specific scenarios in VANET
simulation. We will further improve the VANET
channel model by considering more factors in urban
trafficenvironment.
ACKNOWLEDGMENT
This work is supported financially by the Science and
Technology Key Project of 12thfiveyear plans of
AnhuiProvince(GrantedNo:11010402081).
REFERENCES
A. Goldsmith, Wireless Communications. New York:
CambridgeUniv.Press,2005.
Andr e Cardote, Filipe Neves. A Statistical Channel Model
for Realistic Simulation in VANET. 2012 IEEE Vehicular
NetworkingConference(VNC).
Christoph Sommer, Stefan Joerer, Falko Dressler. On the
ApplicabilityofTwoRayPathLossModelsforVehicular
Network Simulation. 2012 IEEE Vehicular Networking
Conference(VNC).
Dali Wei, Feng Chen and Tongshuang Zhang. Least
SquareSupport Vector Regression based Carfollowing
ModelwithSparseSampleSelection.Proceedingsofthe
8th World Congress on Intelligent Control and
AutomationJuly692010,Jinan,China.
ITUR, Propagation by diffraction, International
Telecommunication Union, Recommendation ITUR
P.52611,Oct.2009.
J. Maurer, T. Fugen, T. Schafer, and W. Wiesbeck. A new
intervehicle communications (ivc) channel model in
Proc. IEEE 60th Vehicular Technology Conference (VTC
2004Fall),vol.1,Sept.2004,pp.913.
J. Turkka, M. Renfors. Path Loss Measurements for a
NonLineofSight MobiletoMobile Environment. ITST
2008. 8th International Conference on ITS
Telecommunications,2008.
www.ijape.org InternationalJournalofAutomationandPowerEngineering(IJAPE)Volume2Issue4,May2013
122
Kunisch.J, Pamp.J. Wideband CartoCar Radio Channel
Measurements and Model at 5.9 GHz. Vehicular
TechnologyConference,2008.VTC2008Fall.
Lin Cheng, Benjamin E. Henty. Mobile VehicletoVehicle
NarrowBand Channel Measurement and
Characterization of the 5.9 GHz Dedicated Short Range
Communication (DSRC) Frequency Band. IEEE Journal
on Selected Areas in Communications, Vol. 25, No. 8,
October2007.
MateBoban,TiagoT.V.Vinhoza,MichelFerreira.Impactof
Vehicles as Obstacles in Vehicular Ad Hoc Networks.
IEEEJournalOnSelectedAreasInCommunications,Vol.
29,NO.1,January2011.
Maria Kihl, Kaan Br, Fredrik Tufvesson. Simulation
Modelling and Analysis of a Realistic Radio Channel
Model for V2V Communications. 2010 International
Congress on Ultra Modern Telecommunications and
ControlSystemsandWorkshops(ICUMT).
PanagiotisPaschalidis,KimMahler.PathlossandMultipath
Power Decay of the Wideband CartoCar Channel at
5.7GHz.VehicularTechnologyConference(VTCSpring),
2011IEEE73rd.
Scott Biddlestone, Keith Redmill. An Integrated 802.11p
WAVE DSRC and Vehicle Trafc Simulator With
Experimentally Validated Urban (LOS and NLOS)
Propagation Models. IEEE Transactions on Intelligent
TransportationSystems,Vol.13,No.4,December2012.
Thomas Mangel, Oliver Klemp. A Validated 5.9 GHz
NonLineOfSight PathLoss and Fading Model for
InterVehicle Communication. 2011 11th International
ConferenceonITSTelecommunications.
T.S.Rappaport,WirelessCommunications:Principlesand
Practice, 2nd ed. Upper Saddle River, New Jersey:
PrenticeHallPTR,2009.
Xinxin Sun, Feng Chen. An Improved Dynamic Load
Balancing Algorithm for Parallel Microscopic Traffic
Simulation. 2012 1ntemational Conference on
Measurement,InformationandControl(MIC).

Potrebbero piacerti anche