Path Planning Algorithm For Extinguishing Forest Fires

JOURNAL OF COMPUTING, VOLUME 4, ISSUE 2, FEBRUARY 2012, ISSN 2151-9617
https://sites.google.com/site/journalofcomputing
WWW.JOURNALOFCOMPUTING.ORG
108
Path Planning Algorithm for Extinguishing

Forest Fires
M.P.Sivaram Kumar, S.Rajasekaran
Abstract One of the major impacts of climatic changes is due to destroying of forest. Destroying of forest takes place in many ways but
the majority of the forest is destroyed due to wild forest fires. In this paper we have presented a path planning algorithm for extinguishing fires
which uses Wireless Sensor and Actor Networks (WSANs) for detecting fires. Since most of the works on forest fires are based on Wireless
Sensor Networks (WSNs) and a collection of work has been done on coverage, message transmission, deployment of nodes, battery power
depletion of sensor nodes in WSNs we focused our work in path planning approach of the Actor to move to the target area where the fire has
occurred and extinguish it. An incremental approach is presented in order to determine the successive moves of the Actor to extinguish fire in
an environment with and without obstacles. This is done by comparing the moves determined with target location readings obtained using
sensors until the Actor reaches the target area to extinguish fires.
Index Terms Forest Fires, Wireless Sensor and Actor Networks, Path Planning
1 INTRODUCTION
Nowadays the likelihood of occurrence of forest fire is weatherconditionslikemistwillhavepoorvisibilityand
steadilygrowingbecauseofchangeinclimateduetocut life on the tower is miserable. In order to surmount the
down of trees in forest for various reasons such as Land disadvantagesfacedbyabovemethodssomevisiontech
for shelter, Land for cultivation and killing of dangerous niques such as Automatic Video Surveillance systems
wildlife animals which pose threat to human lives. The were proposed to monitor small forest. Some of the fire
Forest Survey of India reports that actual forest cover of detection mechanisms such as AVHRR and MRIS which
India is 19.27% of the geographic area corresponding to are based on satellite images are also proposed. But the
63.3 million hectare [1]. It is also estimated that the pro long scan period and low resolution of images reduces
portionofforestareaspronetoforestfiresannuallyrang the effectiveness of satellite based forest fire detection
esfrom33%insomestatestoover90%inothers[1].The mechanismsbecausefiredetectionisverylateandbythat
forestfireseasoninIndiaisusuallyfromFebruarytoMid time fire is detected it may have grown large [4] and
June. An estimated annual economic loss of 400 Crore some of the of resources might have been destroyed. In
rupees is reported on account of forest fires over the ordertoovercomethedifficultiesassociatedwithsatellite
country.Henceitisessentialthatresearchactivitiesmust imagebased mechanisms manyworks havebeenreport
be carried out for forest fire detection and extinguishing ed [5,6] using WSNs because sensory information can
in order to save valuable resources, environment and provide a more comprehensive forest fire monitoring
economyofthecountry.Modelingfiresisusedtounder withafinergrainedspatialandtemporalresolution.Also
stand and predict fire behavior without getting burned sensor nodes can be deployed in regions where there is
[2].InAntiquedaysfiredetectioniscarriedoutvisiblyby nosatellitesignal.Mostoftheworksreportedintheliter
the forest guard who were walking along predefined atureareconcernedwithdetectionoffiresbasedonsimu
pathsalsocalledgroundpatrollingandalsobyforestde lationorusingrealtestbedEnvironments[6].Recentad
partmentpersonnelemployedonthelookouttowersfor vancements in technology have lead tothe emergence of
keeping a vigil in the forest area through binoculars [3]. distributedWirelessSensorandActorNetworks[7].In[8]
Inbothcasesinformationisgiventothecontrolroomand forest fire detection and extinguishing using WSAN was
appropriateactionswillbetaken.Alsoboththemethods proposed. In this paper we have presented a Grid based
are not reliable because the path may get damaged, bad model based on matrix coordinates where each cell of a
grid contains one sensor positioned at the center and a
Path Planning algorithm for theActor to extinguish fire.
M.P.Sivaram kumar, Research Scholar, Department of Computer Science

The remainder of the paper is organized as follows: sec
and Engineering, B S Abdur Rahman University, Vandalur, Chennaition2describesGridbasedmodel.Insection3wepresent
600048.
themethodologyusedforforestfiredetectionandextin
S.Rajasekaran, Professor, Department of Mathematics, B S Abdur
Rahman University, Vandalur, Chennai-600048.
guishing. Section 4 portrays the simulation results while
section5concludesthepaper.

2 GRID BASED MODEL

WSANsrefertoagroupofSensorsandActorsconnected
by wireless medium to perform distributed sensing and
acting tasks. In WSANs the role of sensors is to gather
information from the environment, while actors collect
and process data and perform appropriate actions. The
Notationsandassumptionsusedinthegridbasedmodel
isshownbelow
20.ThePathcalculatedbytheActorisshownusingaLine
incyancolor.
2.1 Notations and Assumptions

1.DefinetheforestareatobeAwherethesensorsare
Initiallydeployed.
2.ConsiderasetofsensorsS={s1,s2,s3sn}arrangedin
sideagridbasedonmatrixcoordinatesinatwodimen
sionalEuclideanregionasshowninFig1.
3.EachSensorisplacedinsidetheforestareaAatcoor
dinate(xi,yi)basedonelementsofamatrixandthe
sensorknowsitsownlocation.
4.Thesensingregionofasensorisassumedtobecircular
withradiusr,whereriscalledassensingrangeof
sensor[9]
5.Thelengthofagridisequaltotwicetheamountof
sensingrange.i.e.2r=LwhereListhelengthofthe
grid[9]
6.Allthecellsinsidethegridareofequalsizeandhave
thesamesensingranger.
7.EachsensorhasanOmnidirectionalantenna.i.e.itcan
perform360degreeobservation.
8.FoursensornodesareplacedatthecornersoftheDo
main.
9.Forestfirepropagatesinalldirectionsatsamespeed.
10.Sincepropagationoffireisassumedtobeuniform
andwehaveconsideredsurfacefiresonly,fireorigin
natedinaregioni/celliisdetectedbysensorifirst.
11.ActorisavailableatthecenteroftheDomainandre
turnstothesameplaceafteritextinguishesfire.
12.Actorcontainstheprocessingunittodecidenextse
quenceofmovestoreachthetargetareaandextin
guishfire.
13.Actorscontainfireextinguishingpowdertosuppress
fire.
14.Actorcanmoveinanyoneoftheeightdirections[10]
numbered1,2,3,4,5,6,7,8fromitscurrentpositionas
showninFig2.
15.Theobstaclepositionsarepredefined.
16.InitialpositionofanActorisrepresentedusinggreen
ovalanditismobile.
17.Obstaclesarerepresentedusingyellowcolorandthe
maximumlength,widthoftheobstacleisequaltothe
sideofsquare.i.e.obstacleisnotgreaterthansingle
squarecell.
18.Obstaclesarestatic.
19.Atwodimensionalarraywillholdavalueof1foran
obstacleinacellwhosecoordinatesarex,y.
i.e.O[x,y]=1
109
(1,1)
(1,2)
(1,n-1)
(1,n)
(2,1)
(2,2)
(2,n-1)
(2,n)
(m-1,1)
(m-1,2)
(m-1,n-1)
(m-1,n)
(m,1
(m,2)
(m,n-1)
(m,n)
Fig 1 Forest Area A divided into cells Based on Matrix

Coordinates
8
5
7
6
Fig 2 Possible directions Actor can Move
3.ForestFireExtinguishingModel
In this work we have explored the use of WSANs in de
tecting and extinguishing forest fires. The automatic for
est fire detection and extinguishing system consists of
nodes deployed deterministically in a forest area and all
thenodesknowtheirlocationbasedoncoordinatevalues
of a matrix. Each node is equipped with a temperature
sensorandanOmnidirectionalantenna.Nodescontinu
ouslymonitortheenvironmenttocheckifthereisfireor
notintheparticularcell.Whenchangeintemperaturei.e.
temperature raises above a certain threshold, is detected
by a particular node they send message packets which
contain location measurements. These packets are re
ceived by one of the corner node. The corner node then
sends the packet to theActor which in turn will process
thepacketwhichcan beusedinreaching thetargetarea
toextinguishfire.
Sincethefocusofthisworkismainlyonforestfiredetec
tion and extinguishing mechanism we assume that mes
sagescommunicatedhasreachedthecorrespondingdes
tinationssafelyandifmorethanonenodeissendingthe
samemessage,themessagesentbythefirstnodeiscon
sidered and rest of them are ignored. The Fundamental
taskstobeconsideredinachievingthisareclassifiedinto
following
a.SensorDeploymentandCoverage
b.CommunicationbetweenNodes
c.PathplanningfortheActor
3.1 Sensor Deployment and Coverage

According to [10] sensors major function is to sense the
environmentforanyincidenceofeventofinterest.There

fore coverage is one of the major concerns in sensor net

works. In [12], Coverage is classified into three classes:
Area coverage, Point Coverage and Barrier coverage.
Since our objective is to maximize the coverage percent
ageinthispaperwefocusedonareacoverage.Thecover
age problem is defined as how to position or deploy the
sensors in a particular Region of Interest (ROI) so that
coveragepercentageismaximizedandcoverageholesare
minimised.The deployment of nodes can be done either
randomlyordeterministically.Inthisworkwehavecon
sidered the deterministic deployment of nodes because
sensor network coverage can be improved by carefully
planningpositionofsensorsintheROIpriortotheirde
ployment [12]. Grid based sensor networks divide the
ROIintosquarecells[8]andsensorscanbeplacedatthe
center of the square cell in order to maximize the cover
age and also the number of sensors required for placing
insidethesquarecellislesserthanthenumberofsensors
required for placing at the intersection of the grids [8].
Henceinthisworkwehaveplacedthesensorsatthecen
ter of the square cell. In case of Grid based deployment,
problem of coverage of sensor field reduces to the prob
lem of coverage of one cell and its neighbor because of
symmetryofcells.Thereforeifwetakeonecellandcalcu
latetheuncoveredareaitisapproximatelyequalto0.86r2
[13] where r is the sensing range. In this work we have
assumedthatfirespreadsatapointandifitoccursinside
theuncoveredareaofsensori,thenitwillbedetectedby
sensorifirstratherthananyothersensorbecausesurface
firespreadsataconstantrateinalldirections[8].
3.2 Communication between Nodes

In this work we have used three nodes (i) sensor nodes
capableofdetectingachangeintemperaturewhichindi
catesthefireoccurrenceanditbroadcaststhemessageto
the neighboring cells (ii) Corner nodes receives the mes
sagesentbythesensornodesandithasgothighcapacity
fortransmissionofmessages(iii)Actornodereceivesthe
information from corner node the coordinates of the cell
wherethefirehasoccurred.Thenwithitsowncoordinate
asstartpositionandreceivedcoordinatesasgoalposition
it plans a path along which it will travel and reach the
goal position and start extinguishing fire. When more
than one corner node is sending the same coordinate in
formation regarding fire it will be discarded. In these
workswehaveassumedsinglepointofoccurrenceoffire
in the entire Domain. Therefore there is no possibility of
different corner nodes sending different coordinates for
fire occurrence. Also we assume that the messages sent
fromonenodetoanothernodewillreachwithoutloss.
3.3 Path Planning for the Actor

The path panning for mobile robots is defined as the
searchforapathwhicharobothastofollowinaprede
finedenvironmentinordertoreachaparticularposition
[14].Pathplanningisconsideredasanoptimizationprob
110
lem of finding a path between start and end points that

should be free of collision and the path obtained is the
shortestone[15].Mobilerobotpathplanningcanbeclas
sified as (i) static or dynamic (ii) local or global and (iii)
complete or heuristic based on environment, algorithm
andcompleteness.Staticpathplanningreferstotheenvi
ronment which does not contain moving objects and ob
stacles.Dynamicpathplanningreferstotheenvironment
which contains moving objects and obstacles. Both static
anddynamicpathplanningcontainsmobilerobot.When
the robot knows the information about the environment
thepathplanningissaidtobeglobal.Onthecontraryif
therobotdoesnotknowtheinformationabouttheenvi
ronmentthenthepathplanningislocal.Inthisworkthe
pathplanningofactortoreachthetargetareaandextin
guish fire is simulated using the actions performed by a
robot.Inourworkwehaveconsideredpathplanningfor
twotypesofenvironment(i)Environmentwithoutobsta
cles.Thistypeofenvironmentissimilartosavannahtype
forest where there are grasslands only available and no
obstacles.(ii)Environment withobstacles.Thisissimilar
toordinaryforestwheretherearetrees,stonesandsome
other static obstacles.The path planning of the Actor is
doneinincrementalstepsbytakingintoconsiderationthe
start position coordinates and goal position coordinates.
Sinceinthisworkweareusingmatrixbasedcoordinates
forthegrid,thedeterminationofdirectionofinwhichthe
actor has to move presented in [16] is modified and it is
shownbelow
S.No
1.
2.
3.
4.
5.
6.
Condition of the coor

dinates
(xsxg =0)and(ysyg<0)
(xsxg>0)and(ysyg)<0)
(xsxg>0)and(ysyg)=0)
(xsxg>0)and(ysyg)>0)
(xsxg=0)and(ysyg)>0)
(xsxg<0)and(ysyg)>0)
7.
8.
(xsxg<0)and(ysyg)=0)
(xsxg<0)and(ysyg)<0)
Direction it has to
move
Right
DiagonalUpRight
Up
DiagonalUpLeft
Left
Diagonal
Down
Right
Down
Diagonal
Down
Left
3.3.1 Path planning Algorithm for Environment

without obstacles
Step1:Readthexgcoordinateandygcoordinatefromthe
messagereceivedfromcornernode.
Step2:SinceweknowthatActorwillalwaysbeavailable
at the center assign the coordinate values of actor to xs
andys.
Step3:Repeatuntil(xs=xgandys=yg)
if((xsxg=0)and(ysyg<0))then
MoveRight
elseif((xsxg>0)and(ysyg)<0))then
MoveDiagonalUpRight

elseif((xsxg>0)and(ysyg)=0))then
MoveUp
elseif((xsxg>0)and(ysyg)>0))then
MoveDiagonalUpLeft
elseif((xsxg=0)and(ysyg)<0))then
MoveLeft
elseif((xsxg<0)and(ysyg)>0))then
MoveDiagonalDownRight
elseif((xsxg<0)and(ysyg)=0))then
MoveDown
elseif((xsxg<0)and(ysyg)<0))then
MoveDiagonalDownLeft
endif
endif
endif
endif
endif
endif
end
Step4:StoptheActorandstartextinguishfire.
3.3.1 Path planning Algorithm for Environment with

obstacles
Step1:Readthexgcoordinateandygcoordinatefromthe
messagereceivedfromcornernode.
Step2:SinceweknowthatActorwillalwaysbeavailable
atthecenterassignthecoordinatevaluesofactor
toxsandys.
Step3:do
{
d=direction(xs,xg,ys,yg)
switch(d)
{
Case1:if(O[xs1,ys+1]!=1)then
MoveDiagonalUpRight
else
Callfreespacesearch(xs,ys,1)
endif
break
Case2:if(O[xs1,ys]!=1)then
MoveUp
else
endif
break
Case3:if(O[xs1,ys1]!=1)then
MoveDiagonalUpLeft
else
endif
break
Case4:if(O[xs,ys1]!=1)then
MoveLeft
else
111
endif
break
Case5:if(O[xs+1,ys1]!=1)then
MoveDiagonalDownRight
else
endif
break
Case6:if(O[xs+1,ys]!=1)then
MoveDown
else
endif
break
Case7:if(O[xs+1,ys+1]!=1)then
MoveDiagonalDownLeft
else
endif
break
Case8:if(O[xs,ys+1]!=1)then
MoveRight
else
endif
break
}
}while((xsxg!=0)and(ysyg!=0))
Step4:StoptheActorandstartextinguishfire.
3.3.2.1Procedure direction
Thisprocedureisusedtodeterminethedirectiontheac
tor has to move based on start and goal coordinates. It
willreturnanintegervaluebasedonthedirectionithas
tomove.
intdirection(xs,xg,ys,yg)
{
if((xsxg>0)and(ysyg)<0))then
return1
elseif((xsxg>0)and(ysyg)=0))then
return2
elseif((xsxg>0)and(ysyg)>0))then
return3
elseif((xsxg=0)and(ysyg)>0))then
return4
elseif((xsxg<0)and(ysyg)>0))then
return5
elseif((xsxg<0)and(ysyg)=0))then
return6
elseif((xsxg<0)and(ysyg)<0))then
return7
elseif((xsxg=0)and(ysyg<0))then
return8
endif
endif

endif
endif
endif
endif
end
}
3.3.2.1Procedure freespacesearch
This procedure is used to determine the cell which does
notcontainanobstacleifthenextmoveoftheactortothe
cell contains an obstacle. It takes three arguments the s
which refers to x coordinate, f which refers to y coordi
nateandeisthedirectionfromthecurrentcellinwhich
checkforobstacleismadeandcellcontainedanobstacle.
Therefore we perform free space search i.e. finding the
next cell which is free without any obstacle in the anti
clockwisedirectionusingtheprocedureshownbelow.
freespacesearch(s,f,e)
{
switch(t)
{
Case1:if(O[xs1,ys+1]!=1)then
xs=xs1
ys=ys+1
Movetoxs,ys
return
break
Case2:if(O[xs1,ys]!=1)then
xs=xs1
Movetoxs,ys
return
break
Case3:if(O[xs1,ys1]!=1)then
xs=xs1
ys=ys1
return
break
Case4:if(O[xs,ys1]!=1)then
ys=ys1
Movetoxs,ys
return
break
Case5:if(O[xs+1,ys1]!=1)then
xs=xs+1
ys=ys1
Movetoxs,ys
return
break
Case6:if(O[xs+1,ys]!=1)then
xs=xs+1
Movetoxs,ys
return
break
Case7:if(O[xs+1,ys+1]!=1)then
112
xs=xs+1
ys=ys+1
Movetoxs,ys
return
break
Case8:if(O[xs,ys+1]!=1)then
ys=ys+1
Movetoxs,ys
return
break
Default:Nopathfortherobottomove
exit
}
}
4. Simulation results
4.1Simulation in an Environment without obstacle
In this work we have proposed a grid containing m x n

cells based on matrix coordinates as a forest domain in
whichfireoccurrenceisautomaticallydetectedbysensors
and extinguished by actors. In the simulation part we
have not shown the corner nodes because we have as
sumed that message sent by sensors to corner node and
from corner node to actor is reached without any loss
since the prime consideration of this work is extinguish
ingofforestfires.Alsowehaveconsideredasinglepoint
ofoccurrenceoffiresincewehaveonlyoneactorforthe
entire forest domain. We have devised a path planning
algorithmfortheactorinordertoextinguishfireinboth
types of environment.i.e. Environment with and without

113
obstacles.Wehaveusedjavaforsimulationpurposes.We wards the target area and also the time taken for the
createda20x20grid.Theactorisrepresentedbyagreen rotationofrobotinordertoperformnextmove.
oval located at the center, the cells containing obstacles
are represented using yellow and the cell where fire oc 6. References
curs is shown in red and a line in cyan color shows the [1]Http://www.fire.unifreiburg.de
[2]QasimSiddque,SurveyofForestFireSimulation,Globaljour
path planning of theActor to travel and reach the target
nalofcomputerscienceandtechnology,pp.137148
area to extinguish fire. To test the effectiveness of the al [3]Pablo.I.Fierens,NumberofWirelesssensorsneededtodetecta
gorithm we have proposed, we created fire in various wildfire,InformalPublication
cells including all the quadrant regions and horizontal [4]M.Hefeeda,M.Bagheri,ForestFireModelingandEarlyDetec
and vertical lines by varying the number of obstacles. In tionusingWirelessSensorNetworksAdHoc&SensorWireless
allthecasesthepathiscreatedusingthealgorithmwith NetworksVol.7,pp.169224
[5]ZhangJG,LiWB,KJM,Forestfiredetectionsystembasedon
out any collision with the obstacle and the actor travels ZigBeewirelesssensornetwork[J],JournalofBeijingForestry
throughthepathtoextinguishfire.Thesimulationresults University,v29,n4,July,2007,pp.4145
areshownbelow.
[6]B.Kosucu,K.Irgan,G.Kucuk,S.Baydere,FireSenseTB:Awire
4.2 Simulation in an Environment with obstacle
5. Conclusion
Inthispaperwehavepresentedapathplanningalgo
rithmfortheactortomovetothetargetareatoextin
guish fire. In this work we have assumed that robot
can move in all directions without any rotation of ro
bot.i.e.iftherobotisfacingtowardseastandthenext
movement it has to make is towards north then robot
hastodosomeadjustmentsothatitcanturntowards
north first and then move towards the cell. We have
considered only 9 obstacles for 20 x 20 environments.
Thissimulationwillworkforanynumberofobstacles
placed in any position. In this work we have consid
ered only static obstacles but in future we have plans
to incorporate as many dynamic obstacles as possible
andalsoconsidertimetakenfortherobottomoveto
lesssensornetworksTestBedforForestFireDetection,pp.1173
1177
[7]IanF.Akyildiz,IsmailH.Kasimoglu,WirelessSensorandActor
Networks:researchChallenges,MobileAdhocNetworks,Vol.no
2,pp.351367
[8]M.P.SivaramKumar,S.Rajasekaran,DetectionandExtinguish
ingForestFiresusingWirelessSensorandActorNetworks,
InternationalJournalofComputerApplication,Vol.24(1),June
2011,pp.3135.
[9]BLiu,DTowsley,Onthecoverageanddetectabilityoflarge
scalewirelesssensornetworksproceedingsoftheworkshopon
ModelingandoptimizationinMobileAdhocandwirelessnet
works(Wiopr03),Mar2003
[10]MSedaandTBrezina,RobotMotionPlanninginEightDirec
tions,ProceedingsoftheWorldCongressonEngineeringand
ComputerScience2009VolIIWCECS2009,October2009,San
Francisco,USA
[11]N.A.AbAziz,K.AbAzizandW.ZWanIsmail,CoverageStrate
giesforWirelessSensorNetworks,WorldAcademyofScience,
EngineeringandTechnology502009,pp.145150
[12]CardeiMandWuJ,CoverageinWirelessSensorNetworks
InformalPublication
[13]Jiehuichen,MBSalimandMMatsumoto,Asinglemobiletar
gettrackinginvoronoibasedclusteredwirelesssensornetwork,
JournalofInformationprocessingsystemsVol.7,March2011,
pp.1728
[14]BuniyaminN,WanNgahWAJ,SariffN,MohammadZ,A
SimpleLocalPathPlanningAlgorithmforAutonomousMobile
robots,InternationalJournalofsystemsapplications,
Engineering&developmentissue2,volume6,pp151159,2011
[15]A.howard,M.J.MatariandG.S.Sukhatme,AnIncrementalself
deploymentalgorithmforMobilesensorNetworks,autonomous
robots,specialissueonIntelligentEmbeddedsystems,13(2),
September2002,pp.113126.
[16]TingKaiWang,QuanDang,PeiYuanPan,PathPlanningAp
proachinUnknownEnvironment,InternationalJournalof
AutomationandComputing,7(3),August2010,pp.310316.

Path Planning Algorithm For Extinguishing Forest Fires

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Path Planning Algorithm For Extinguishing Forest Fires

Hochgeladen von

Copyright:

Verfügbare Formate

JOURNAL OF COMPUTING, VOLUME 4, ISSUE 2, FEBRUARY 2012, ISSN 2151-9617