Usage of mobile elements in internet of things environment for data aggregation in wireless sensor networks R

(1)

ContentslistsavailableatScienceDirect

Computers and Electrical Engineering

journalhomepage:www.elsevier.com/locate/compeleceng

Usage of mobile elements in internet of things environment for data aggregation in wireless sensor networks ^R

Hanady M. Abdulsalam

^∗

, Bader A. Ali, Eman AlRoumi

Department of Information Science, Kuwait University, Kuwait

a r t i c l e i n f o

Article history:

Received 29 May 2017 Revised 12 December 2017 Accepted 13 December 2017 Available online xxx Keywords:

Wireless sensor networks Data aggregation Data streams Internet of things (IoT)

a b s t r a c t

Themain issueto beaddressedin WirelessSensornetworks (WSN)applications isthe limitedlifetimeofsensorsandshortcommunicationrange.Professionaldataaggregation techniquesare, therefore,needed. Inthispaper, we considerthe problemof increasing theWSNlifetimeusing acluster-baseddataaggregation algorithm.We proposeanovel methodintacklingtheproblem.WeuseMobileElements(ME)inInternetofThings(IoT) environmenttoactasClusterHeadsinacluster-basedaggregationalgorithm.Webelieve thatutilizingtheIoTtechnologybymixingitwiththeWSNtechnologyleadstogoodre- sults.Ourexperimentsshowanimpressiveextendthenetworklifetime,whilenoteffecting thequalityofdatagathering.

1. Introduction

WirelessSensorNetworks(WSN)arewirelessnetworksthatconsistofalargenumberofsensingdevices(sensors)that areeitherdistributedamonggeographicalareassuchasenvironmentalmonitoringandflooddetectionorattachedtospe- cificobjectssuchaspatientsinhealth-careandcarsintrafficcontrolapplications.AtypicalWSNcollectsdatarecordsfrom thesensorsandsendsthemtobasestationstobeprocessedandanalyzeddependingontheapplicationthatitisusedfor.

Decisionsare,then,takenbasedontheanalysis.Examplesofthedatathatcanbecollectedincludetemperature,humidity, lightconditions,seismicactivities,etc.

DataaggregationinWSN forman ongoingactiveresearcharea duetoitsimportance insolvingthemaindrawbacksof usingWSNs,namely,thelimitedlifetimeforbatterypoweredsensors,andtheshortcommunicationrangeofthesensors.A numberofdataaggregationmethodsappearintheliterature,suchasLEACH,DD,FEDA,TAG,OCABTR,CLUDDA,SUMAC,etc.

[1–8].Many oftheexisting aggregationalgorithms forWSNsare designedto workwithcluster-based routing algorithms [1–6],whilesome othersarebasedonotherwaysofrouting[7,8]suchasdirectconnectionbetweensensorsandthebase station.

Inthispaper,we only considerthe cluster-basedrouting algorithms. The cluster-based routing algorithms formatwo layersendingprocedure.Theygroupnearbysensorstogethertoformanumberofclusters.Eachclusterhasonerepresen- tativesensor calledClusterHead (CH).CHsare requiredto collect datafrom other sensors intheir cluster andsend the aggregateddatatothebasestation.ThedecisionofchoosingCHsincluster-basedaggregationalgorithmscanbeeithercen- tralized;decidedbythebasestation,ordistributed;basedonthedecisionofthesensors.Thekeyfeatureofcluster-based

R Reviews processed and recommended for publication to the Editor-in-Chief by Associate Editor Dr. V. Sai.

∗ Corresponding author.

E-mail address: [email protected] (H.M. Abdulsalam).

(2)

2 H.M. Abdulsalam et al. / Computers and Electrical Engineering 0 0 0 (2017) 1–19

ARTICLE IN PRESS

JID:CAEE [m3Gsc;December19,2017;11:59]

algorithmsinincreasingthenetworklife-timeisthat,becausenotallsensorsinteractwiththebasestation,moreenergyis savedandthenetworklifetimeisextended.

Recently,severalWSNarchitecturesbasedonmobileelements(MEs)havebeenproposed[9–13].MobileElementscanbe anyexternaldevicethatispassingthroughthenetworkwithsensingandsendingcapabilities.Forcluster-based networks, theroleofMEsineacharchitectureisdifferent[14].MEscanbe:

Regularsensornodesthataresourcesofdata,responsibleforsensingthedatarecordsandsendingthemtoCHs, Basestationsthatarethedestinationsofdataand,areusuallytheprocessingelementsinthenetwork,and/or Clusterheadsthataggregatedataasintermediatedatacollectorsorgatewaystobasestations.

WeconsiderthescenariowhereMEstakeonlytheroleofclusterheads.However,weassumethattheMEsthatareused intransferringthedataarenotWSN-registeredmobilesensors.Theyare,instead,anonymousMEsthatarepassingthrough thenetwork,thengotdiscoveredbytheInternetofThings(IoT)technology,andconnectedtothewirelessnetwork.Internet of Things (IoT) [15] is a recent technology and an active research area in the ﬁeld of objects communication anddata transmission.Itisatechnologythatconnectsdifferentobjects,suchassmart-phones,sensors,andpeopletotheInternetin ordertoallowaccessibilitytothedata.

We propose an energy-efficient data-aggregationtechnique forWSN using MEs in an IoT environment. The proposed solutionisacluster-basedalgorithmsuchthattheMEsactasCHs.WetakeintoconsiderationthefrequencyatwhichMEs enterthenetworkaswellastheirspacedistributionamongthenetworkarea.DependingontheflowofMEsintotheWSNs, wedefinefourpossiblescenarios:

Scenario0is thebasicscenariowhereit isassumedthat theMEscontinuously ﬂow intothenetwork, anduniformly distributedwhiletravelingthroughthenetworkarea.Clearly, suchsituationisbasicandidealsinceinrealapplica- tions it can not be guaranteed that there willbe MEs passing through the network all the time andyet traveling uniformlythrough it. The onlynew issuefor Scenario0 is that theCHs arenot regular sensorsanymore, instead, theyareMEsthatare interchangeablypassing throughthenetworkwithhighpowercapabilitieswhencomparedto regularsensors.Hence,thebatterypowerlimitisnomoreanissuelikewhenusingregularsensor.

Scenario1isthescenariothatintroducestheconceptofnonuniformdistributionofMEsamongtheareaofWSNwhile frequently having them presented in the network. In this scenario, the MEs do not uniformly travel through the network.Weassume havingthem biasedtowardsone sideofthenetwork area.Themainissuetobeaddressedfor thisscenariois,therefore,thecoverageoftheareaswithnoMEspassingthroughthem.

Scenario2isthecasethatassumestherearetimeperiodsofthenetworklifetimeduringwhichtherearefeworprob- ablyno MEs passingthrough thenetwork. Thiscasedeﬁnesthenonuniform entrancefrequency oftheMEs tothe network.ThecoverageofthesetimeperiodswithnoMEsaretheconcernofthisscenario.

Scenario3isthemostgeneralcase,whereitassumesamixturebetweenalltheabovescenariosthroughdifferenttime periods,suchthateachtimeperiodofthisscenariocanbeeitherScenario0,1,or2,oranoverlapoftwoscenarios.

Allofthepreviouslymentionedscenariosare,therefore,specialcasesofScenario3atsometimeperiod.

Tohandlethedifferentscenarioswhilenotaffectingthedataaggregationaccuracy,wefurtherimproveourproposalinto ahybridsolutionthatswitchestheaggregationfromMEdependenttechnologytotheregularWSNtechnologybasedonthe unavailabilityoftheMEstoactasCHsintothenetwork.Thisensuresthat theaggregationisindependentofthepresence oftheMEs.

Wesimulateouralgorithmusingthedifferentabovementionedscenarios.Ourresultsshowthatthenetworklifetimeis improved,whilenotsigniﬁcantlyaffectingthedataaggregationaccuracy.

Theremainder ofthepaperisorganizedasfollows:Section 2statesbackgroundandrelatedwork.Section 3proposes ourbasicandhybriddataaggregationalgorithmusingMEsinIoTenvironment.Section4explainstheimplementationand experimentalsettings.Section5statesthediscussionandresults.Finally,Section6drawsourconclusionsandfuturework.

2. Backgroundandrelatedwork 2.1. Background

2.1.1. Internetofthings

TheInternet ofThings(IoT) computingmodelhasgainedalotofpopularityinthelastdecade.Recently,an increasing numberof IoTprojects are beingdonein differentareaslikeagriculture, environmentalmonitoring, andsecuritysurveil- lance.Initially,thetermIoTusedtorefertotheinteroperabilityofuniquelyidentifiableobjectswithradio-frequencyiden- tification (RFID)technology[15].Later,thedefinitionoftheIoT hasexpandedtorefer toanetwork ofinterconnected objects/devices such asRFID tags, sensors, actuators, smart phones,and singleboard computers. These objects are able to collectdatafromitsenvironmentthroughsensors,interactwiththephysicalworldbyprocessingandapplyingactions,and establishcommunicationbyutilizingtheexistingcommunicationprotocolstandards.InanIoTenvironment,theobjectsare expectedtocooperatetoreachcommongoals[16].

AnumberofunderlyingkeytechnologiesformthefoundationonwhichtheIoTrelies[17,18].Oneofthemaintechnolo- giesis RFIDwhich allows chips totransmit identiﬁcation informationthrough wireless communication. RFIDs havebeen

(3)

widelyusedinlogistics,andtrackingapplications.Anotherkeytechnologyisthewirelesssensornetworks(WSNs)utilizing large number of intelligent sensors to collect, aggregate, anddisseminate importantinformation from different environments such asenvironmentalmonitoring,industrial monitoring,traﬃc monitoring, andothers.The differentcommunica- tion/Internetprotocols(i.e.,TCP/IP,WiFi,Bluetooth,Zigbee, etc)provides essentialsupportfortheIoT byenablingusersto controlthe objectsandobjectstocommunicate witheachother.In addition,othertechnologieslikesmart devices,social networksandcloudcomputingarebeingusedtosupporttheIoT.

ThereareseveralapplicationdomainsinwhichIoTtechnologyisbeingdeveloped[17,19,20].Oneofthemaindomainsin whichIoTisbeingusedismobileIoTapplications.Theseapplicationsareconcernedwithmonitoringandsensingofdifferent environments implemented through traditionalWSNs liketraﬃc monitoring,environmental monitoring,and logisitics. In monitoringapplications,gettingaccuratetimely informationisofimportance.UtilizingIoT technologybyrecruiting smart mobileobjectsequippedwithsensorstoparticipateinsensingtaskscangreatlyimprovetheeﬃciency,accuracy,coverage, andavailabilityoftraditionalWSNsaswellasincreasethelifespanofWSNs.

2.2.Relatedwork

Recentwork onWSN cluster-basedaggregationinvolves usingmobilecluster headsinstead ofregularstill sensors [9–

13].Theidea behindthat istoreducethedistancebetweentheCHandthebasestation,hence,lessenergyisconsumed.

Tothebestofourknowledge,noneoftherecentresearchintheareaconsidermixingtheIoTandtheWSNtechnologiesin aggregatingthedata,suchthatexternalarbitraryMEsareCHsonlyandallothersensingdevicesareregularsensors.

Banerjeeetal.proposeaschemeforenhancingthe networklifetimeusingmultipleenergyrichnodesmobile CHsthat canmoveintheWSNinacontrollablemanner[9].ThegoalistoenhancethelifetimeoftheWSNbyproposingevent-driven mobilitystrategiesforCHs.Oncetheclusterformationstepiscompleted,theyarekeptﬁxed,andonlytheCHsmoveinthe directionoftheevent,keepingtheother membersoftheclusterstatic. TheCHcontrollablymovestowardtheenergy-rich sensorsortheeventarea,offeringthebeneﬁtsofmaintainingtheremainingenergymoreevenly,oreliminating multihop transmission.

MaandYang [10]haveproposedan algorithmthat worksonpositioning ofmobilecluster headsandbalancingtraﬃc loadin sensor network that consistsof static andmobile nodes. Theystate that the location ofthe cluster head in the clusterednetworkcan signiﬁcantlyaffectthenetwork lifetime. Movingclusterheads tobetter locationnetwork loadcan, therefore,balancetheloadandprolongthenetworklifetimeaccordingly.

Kunmaretal.[12]enhancetheLEACH-Mobiletechnique[11],whichsupports themobilityofallsensors includingthe CHs.WhileLEACH-Mobileelectsanodewithlessmobilitythanitsneighboursasclusterhead,theEnhancedLEACH-Mobile protocolisbased ona mobilitymetric ”remoteness” forcluster headelection, such that mobilitymeasureshould have a linearrelationship withlinkchangerate.Thisensureshighsuccessrateindatatransferbetweentheclusterheadandthe collectornodeseventhoughnodesaremoving.

LEACH-MAE[13]improvestheLEACHprotocoltosupportmobilityalongwithanewaverageenergybasedCHselection techniquewhichisoptimizedforthemobilenodes.ThisproposedmodiﬁcationismadeonthebasisofCHselectionalgo- rithmtoensurethatpowerresourceisequallydistributedamongthesensornodesandeverysensornodehasanabilityto becomeclusterhead.

3. DataaggregationinWSNsusingmobileelements(MEs)inIoTenvironment

Asit hasbeenmentioned earlier,themain issuetobe addressedinWSNsenvironments isthe limitedbatterypower ofthesensors.Thepowerdropsasthesensorssenddata.Themainparameterthateffectsthepowerdropisthedistance ofsendingthe data.Manyaggregationtechniques do,therefore,useclusterstolimit thedistanceof sendingthe data,so that regular sensors send the data to the cluster heads, while only the cluster heads send the data further to the base stations.Hence,morepowerissaved.The CHsrole is,then,rotatedamongdifferentsensors eitherrandomlyorbasedon somemetrics,suchastheremainingenergyorthedistancetobasestation,orprobablyonlyreplacedwhentheyareoutof power.

3.1. Basicsolutiondetails

Wepropose asolution ofhavingthe CHsasexternal MEs thatpass throughthe network.Foreach round,asMEs are passing through the network, they get discovered by the base stationthrough IoT environment,then their locationsare publishedtoall thesensors inthenetwork aftertakingtheir permissionto participateinthenetwork andact ascluster heads.PermissionsaretakentoensurethatthemainfunctionoftheMEsdonotcollapse.TheMEs,then,collectdatafrom the sensors through IoT, and ﬁnally send the collecteddata to the base stationof the WSN. A ME can be a car sensor, amobile phone sensor,a smart watch sensor, etc.Each roundof datacollection can bebasically thoughtofhaving four phases:

MEdiscoverywhereeachMEisdiscoveredthroughthebasestation

MElocationbroadcastwherethebasestationbroadcaststhelocationofeachMEtothesensors

(4)

ARTICLE IN PRESS

Fig. 1. General solution.

MEdatacollectionwhereeachMEreceivesthedatarecordsfromitsclosebysensors BasestationdatacollectionwhereeachMEsendsthecollecteddatatothebasestation

Fig. 1 shows the general view of the proposed solution. This solution can only handle Scenario 0- described in the introductionsection-whereitisassumedthatMEsarealwayspresentedanduniformlyscatteredamongthenetworkarea.

thenextsubsectionproposesamodiﬁcationtohandleScenarios1through3.

3.2. Hybridsolutiondetails

WefurtherextendoursolutiontohandlethecasesofScenarios1through3,whereMEsarenot uniformlydistributed amongthenetworkareaand/orMEsarenotpresentedduringsometimeperiods.

Weﬁrstproposeameasuretocheckwethertherearesensorsthatarenotsendingdatarecordsforanumberofrounds because noMEs have passedthrough them forsome time period.Reasons of not having MEspassing through them are eitherbecauseoflackofMEsforsometimeperiodsorthenonuniformdistributionofMEsthroughthenetworkarea.

Assumethat thenetworkisdividedintoequalnumberofsquarezonesnz.Eachzonez_i where0≤i≤nz coverspartof thenetwork.Ifatanyroundr_j,where0≤j≤R,andRisthetotalnumberofrounds,a sensorsz_i thatislocatedinzonez_i sendsdatatoaME,thezonez_iissaidtobeacoveredzoneduringroundr_j.Otherwise,zonez_iissaidtobeanuncovered zoneduringr_j.Ifazonez_istaysuncoveredforcnumbersofrounds,thenthiszoneshouldalternatinglyswitchtoaregular cluster-based aggregation algorithmwhere thezone is considered a cluster, anda CH iselectedbased on themaximum remaining energyofthesensorsinthiscluster.Membersofthecluster, then,senddatatotheelectedCHsensor,andthe CHforwardsthedatatothebasestationusingtheregularsettingoftheWSN.Fig.2showsthehybridsolution.

3.3. Solutionalgorithm

Basedontheaboveexplanation,thestepsofperformingtheproposedsolutioncanbesummarizedinAlgorithm1. 3.4. Systemarchitecture

The proposedsolution canbe appliedusingthe architectureinFig. 3.Anexisting WSNis builtontop ofanIoT environment.TheMEsthatare mentionedinthesolutionare assumedtobe partoftheIoTenvironment.The interactionand communicationbetweenthedifferentcomponentsofthesystemareexplainedindetailsinthenextsubsections.

3.4.1. Interaction

Anumberofinteractionstakeplacebetweenthecomponentsofthesystem. FromFig.3,thestaticsensors intheWSN areresponsibleforcollectingdatafromthesensedﬁeld.Eachround,thesensorssendthesenseddatatotheclosestMEthat

(5)

Fig. 2. Hybrid solution.

Fig. 3. System Architecture.

ispassingthroughtheareaoftheWSN.TheMEsrelaytheaggregateddatatothebasestation.TheMEspassingthroughthe WSNarediscoveredandidentiﬁedbythebasestationineachround.Finally,thebasestationisequippedwithanInternet gatewaythroughwhichthedatarecordsaresenttothebackendserversorthecloudforprocessingandstorage.

3.4.2. Communication

Thecommunicationbetweenthedifferentcomponentsofthesystemisachievedthroughanumberofprotocols.Most oftheexistingwireless sensorsnetworkscommunicateusingIEEE802.15.4/ZigBeestandards[21]torelaydatatothebase station.Recently,wireless sensorsequippedwithothercommunicationprotocolslikeWiFiandBluetoothlowenergy(BLE) are startingto emerge [22,23]. BluetoothLow-Energy (BLE)uses a short rangeradio andminimum amountof power to operate.Its coverage rangeis around (100m) andhavea very shortlatency [24]. BLEcan be operated ata transmission powerbetween0.01mWto10mW.ComparedtoZigBee, BLEismoreeﬃcientintermsofenergyconsumption [24]which makesitagoodcandidateforWSNs.Inbothcases,ithasbeenshownthattheenergyconsumptionfromemployingWiFi andBLEisvery lowleavingthesensorswithalong lifespan.Inaddition,WSNsareproved tohavemoreﬂexibilityfrom utilizingthesemainstreamcommunicationtechnologiestoachieveeasierinteractionwithexternalelements.

(6)

ARTICLE IN PRESS

Fig. 4. The different steps of the proposed data aggregation solution.

TheMEsneedtocommunicatewiththewirelesssensorstocollectthesenseddataandatthesametimeneedtocom- municate withthe base stationto relay the collected data.The MEs that we consider inour system like smartphones, smartcars, andsingleboardcomputers(e.g.,Arduino,Raspberryetc.)are assumedto havesimpleprocessingcapabilities, moderatestorage,andcommunicationcapabilitieslikeWiFi/Bluetooth.Therefore,theMEsusetheavailableWiFi/Bluetooth capabilitiestocommunicatewiththesensorsandthebasestation.Atthebasestationend,awirelessinternetgatewaywith widecoverageissetuptoallowthedatatobecollectedfromthemobileelementsandlaterpushedtotheserversorcloud.

3.4.3. Architectureexample

Weexplainthedifferentstagesofthesystemthroughanexample.Figs.4athoughdshowthedifferentstagesofinter- actionbetweenthecomponentsofthesystem.

Inthesystem,MEspassthroughvariouspartsofsensedfieldatdifferenttimes.Atthebeginningofeachdataaggregation round,Fig.4a,thebasestationdiscoversthepositionoftheMEthatareavailablewithin thesensingfield.Thisisdoneby sendingabroadcastmessage(beacon)usingWiFioverthesensingfield.TheMEthatiswillingtoparticipateinthecurrent roundreply backwithan acknowledgemessageidentifyingtheir GPS positionwithin thefield.Inthe nextstep,thebase stationannouncesthepositionoftheparticipatingMEtoallthestaticsensorsinthefieldasshowninFig.4b.Usingthelist ofMEpositions,eachsensordeterminesitsclosestMEbyapplyingasimplecomputation.OncetheclosestMEisidentified, thesensorcommunicateswiththeMEoverBLEandsendsthesenseddataasitspresentedinFig.4c.Atthisstageallthe MEelementshavecollectedthedatafromthestaticsensors.Inthefinalstep,demonstratedinFig.4d,themobileelements relaythedatatothebasestationandthebasestationlatersendsthedatathroughthegatewaytothededicateserversor cloudserviceforfurtherprocessing.

4. Implementationandexperimentalsettings

WesimulateouralgorithmusingC-LanguageandtestitonScenarios0to3.WealsosimulateLEACH[1]asbeingwell- knowncluster-based algorithm tocompareit withourproposed solution.Foroursimulation,we assume thatall sensors candirectlyreacheachothersandreachthebasestation.Hence,nospeciﬁcroutingalgorithmisused.Wealsoassumethat

(7)

Fig. 5. Network structure.

oncethedataissentthenitisreceivedattheotherendsincecalculatingthesuccessrateofdatadeliveryisoutthescope ofthepaper.

4.1. Testingcriteria

Thecriteriabywhichweevaluateourworkare:

- Firstnodedies,lastnodedies,andaveragesensorlifetime:recordstheroundsatwhichtheﬁrstsensorinthenetworkdies andthelast sensorinthe networkdies,andcomputestheaverage oflifetimesforsensorsinnumberofrounds. Note that thenetwork lifetimecanbe adjusted accordingto theobjectives ofthenetwork.Ifthe objectivesofthe network statethatthenetworkisonlyconsideredalivewhenhavingatleastualivesensors,thenthelastnodediesissettobe whenusensorsaredead.

- Numberofalivesensors:recordsthenumberofsensorsthatarestillaliveateachround.

- Remainingenergy:calculatesthetotalremainingenergyforthenetworkforeachround.

- Networkcoverage:showshowthenetworkzonesarecoveredintermofsendingsensorsduringeachround.

Resultsforall criteriaare averagesover20runs,whilerandomlyselectingCHs/sendingsensorsforeachrun. Notethat eachresultrepresentsafullcompleterunforthesimulation,whichmeanseachrunissettocontinuerunninguntilthelast nodedies,andthatexplainsthedifferentnumberofroundsresultingfromeachrun.

4.2.Simulationsettings

4.2.1. Parameterssettings

Wetestthealgorithmsusinganinitialnumberofalivesensorsn=100,eachwithasendingrangeofr=10meters.We useanetworkofsize50×50meters,withabasestationlocatedatpoint[50,100].Thenetworkwebaseourtestingonis showninFig.5.

Thechosen networksizeandnumberofsensorsarewidelyusedintheareaofaggregationinWSN[1,25].Accordingto standards,thecommunicationrangeofsensorscangrowupto100mforoutdoorsensorsand20to30mforindoorsensors dependingontheusedsensors.Weuseashortsendingrangetobeconsideredastheworstcase.Basedontheapplication, higherrangesensorscanbeusedandthenetworksizecanaccordinglybeincreasedaswell.

Weempiricallysetthesizeofthezonestoz=10×10=100m²,assumingit’s areasonabledistancetobe coveredby walkingpeople, whichareourslowestassumedMEsasitwillbeexplainedinSection 4.2.2.Wealsoempiricallytest two

(8)

ARTICLE IN PRESS

Fig. 6. First node dies, last node dies, and average lifetime for LEACH, basic solution of Scenario 0, and hybrid solution for Scenario 0.

Table 1

Radio characteristics.

Operation Energy dissipation

Transmitter/Receiver electronics e elec 50 nJ/bit Transmit ampliﬁer amp 100 pJ/bit/m ²

valuesofc,whichisthenumberofnonsendingroundsatwhichthehybridsolutiongetsactivated,suchthatc=5and10 rounds.Eachroundissettobeoneminute.Thissettingcanbechangedbasedontheapplication.

TheenergycalculationsarebasedontheenergysettingsbyLEACH[1].Theinitialenergye_initialofeachsensorissetto 1J.Thesensors’energyconsumptionfortransimiting/receivingdatainourworkisbasedontheradiomodelandequations showninTable1.

Fortransmittingamessage,

e_T

(

^k,d

)

=e_elec∗k+

^amp∗k∗d², andforreceivingamessage,

eR

(

^k

)

=eelec∗k,

wheredisthedistanceofsendingthemessage,andkisthesizeofthetransmitted/receivedmessage.Wesetk,thesizeof transmittedmessageto2000bits.

4.2.2. Mobileelementssettings

Forthesakeofreality,wedefinethreetypesofMEswithspeeds1,2,and5,wherethespeedunitisjustastepintothe network.WeassumethatanMEwithspeed1wouldbeawalkingperson,whiletheMEswithspeeds2and5arebicycle andcarswithlowspeedsrespectively.Wedonotconsiderthehighspeedelementsinourwork.Wealsoassumewhenan MEisenteringthenetwork thenitcanbeofanytypeofthethreedefinedMEswithequalprobability.Inaddition,when anMEentersthenetworkthenitstraveldirectionischosenrandomlywiththepossibilitytotravelvertically,horizontally, ordiagonally.IftwoMEscollide,theneachMErandomlychoosesanotherdirectionandcontinueitstravel.Notethatthisis consideredtobe theworst casescenariosinceinrealitythisproposedsystemwouldbe practicallyappliedonroadmaps, whichhasdefinedpathstowalkthroughforbothautomobilesandpedestrians.

(9)

4.2.3. Scenariossettings

TheassumptionsonwhichwebaseoursimulationforScenarios0through3are:

Scenario0Isthebasicscenariowhichassumesthefollowing:

- Initially,5MEsenterthenetwork.

- Foreachsubsequentround,theprobabilityofhavingnewMEstoenterthenetworkis10%

- ThemaximumnumberofMEstoenterthenetworkineachroundis5MEs

- EachMEentersthenetworkuniformlyrandomlyfromanyofthefournetworkborders - Ateachround,theremustbeatleast5MEsinthenetwork

Scenario1IsthesameasScenario0,withadifferenceinthespacedistributionofMEsinthenetworksuchthat MEs enterthenetworkwith70%probability fromthetopborderwhileenteringfromtheother threeborderswithequal probabilityof10%each.

Scenario2Dividesthe24h dayinto twotimesperiods torepresentthedayandnighttime periods.Since theround issetto beone minute, the daytimeperiod issetto be 16h length(from round1 to960),while thenight time periodissettobe8hlength(fromround961to1440).Forthedaytimeperiod,thesettingsareexactlythesameof Scenario0,whereasforthenighttimeperiod,thesettingsarechangedto:

- Foreachround,theprobabilityofhavingnewMEstoenterthenetworkis3%.

- ThemaximumnumberofMEstoenterthenetworkateachroundis2.

- At each round,thereis nolimit on theminimumnumber ofMEs inthenetwork. Hence, theremight besome roundswithnoMEsaround.

Scenario3Isthecasewheredifferentscenariosoverlapinarealisticmanner.Wedividethe24hdaysuchthat:

- Scenario0isappliedforrounds1to1250 - Scenario1takesplaceforrounds1251to1440 Whileatthesametimethefollowingholds:

- Scenario0isappliedforrounds0to960

- Scenario2takesplaceassumingthenightsettingsfromround961to1440

Thisdivision leads to havingScenario 0 fromround 1 to960. Then having an overlap ofScenario 0 andScenario 2fromround 961 toround 1250,which meanslessMEs are entering thenetwork uniformlyfromall sides ofthe network.Finally,havingan overlapofScenario1andScenario2fromround1251toround1440,whichmeansless MEsareenteringthenetworknon-uniformlyfromthenetworksides.

5. Resultsanddiscussions

5.1. Firstnodedies,lastnodedies,andaveragelifetime

Figs.6–9representthenumberofroundsatwhichtheﬁrstnodedies,lastnodedies,andshowtheaveragelifetimeof sensorsforScenarios0through 3respectivelywhenappliedasbasicorhybridwithc =5 and10.Allﬁgures includethe LEACHasatypicalbasealgorithmtocomparewith.

ThefiguresshowthatthebasicsolutionextremelyextendsLEACHintermsofthefirstnodedies,thelastnodesdies,and theaveragelifetimeforallscenarios.Thisisbecause theclusterheadrole isnotassignedtoanyregular sensoranymore, instead,theMEsentirelyhandletherole ofclusterhead.Asforthedifferencebetweenthebasicandthehybridalgorithm fordifferentscenarios,itsclearthat increasingthevalueofc to10giveshigherlifetimesforfirstnodediesandlastnode dies,andhigheraveragelifetimethanforthe valueofc=5.Some datarecordsmight,however,bemissedforc=10,so usingthevalueofc=10mightnotbesuitableforsensitiveapplications.

Itcanalsobe noticedthat forScenario0,thedifference betweenthe lastnode diesofthe basicandhybrid isaround 7000rounds whichisabout25% decreaseofthelifetimefromthebasicalgorithmofScenario0,whilethedifference be- tweenthelast nodedies ofthebasicandhybridforother scenariosrangesfrom16,000to 22,000roundswhichisabout 40–50%decrease ofthenetwork lifetimefromthebasicsolution. Whereasthedifferencesinaverage lifetimerangefrom 15%to30% decrease inthe networklifetime fromthebasic solution.Thiscan be explainedasfollows: since theMEs in Scenario0are uniformlydistributedintermsofarea andtime,the basicsolutionperforms wellasthe entirenetwork is coveredmostlybyonlyMEswithouttheneedtoswitch tohybridregularly. Whilefortheother scenarios,thereisalways aplaceoratimeperiodthathasthenetworkuncoveredbyMEs,andsoneedstoswitchtothehybridsolutionalmostreg- ularly.MEs,however,stillworkasCHsmostofthetime,whichcanbeshownfromtheextendinresultswhencomparing thehybridsolutiontoLEACH.NotealsothattheﬁrstnodediesforScenario1ofthehybridsolutionwithc=10iscloseto theﬁrstnodediesofthebasicsolution,sinceforScenario1,therearesomeareasthatswitchtohybridmoreregularthan others,whichmaketheCHsroleassignedtosomesensorsmorethanothers,andhencethesesensorsdiequickly.Whereas forScenarios2 and3, therearetimes atwhich almostall thenetwork hasnoMEs andsothe CHrole isrotatedamong differentsensors,andthereforetheenergyisbetterpreserved.

(10)

ARTICLE IN PRESS

Fig. 7. First node dies, last node dies, and average lifetime for LEACH, basic solution of Scenario 1, and hybrid solution for Scenario 1.

Fig. 8. First node dies, last node dies, and average lifetime for LEACH, basic solution of Scenario 2, and hybrid solution for Scenario 2..

(11)

Fig. 9. First node dies, last node dies, and average lifetime for for LEACH, basic solution of Scenario 3, and hybrid solution for Scenario 3.

Fig. 10. Number of alive sensors versus rounds for LEACH, basic solution of Scenario 0, and hybrid solution for Scenario 0.

(12)

ARTICLE IN PRESS

5.2. Numberofalivesensors

Figs. 10–13 demonstratethe number ofalive sensors versus the numberof rounds forLEACH as a basealgorithm to comparewithScenarios0through3respectivelywhenappliedasbasicorhybridwithc=5and10.

Theﬁgures clearlyindicatethatall basicandhybridalgorithmsextendthenetworklifetimewhencomparedtoLEACH.

When observingthenature ofthegraph ofLEACHandthe basicalgorithmofall scenarios, wecan seethat thenetwork isperforming withone ortwosensorsforaround3000to5000rounds attheendofthenetwork lifetime.Theserounds areconsideredtobeuselesshavingonly1%to2%ofthedatasendingsensors.Whileforthehybridsolution,althoughthe networklifetimeisreducedwhencomparedtothebasicsolution,thenetworkbehaviourisbetterhavingalmostallsensors aliveuntilabout75%to80%ofthetotalnetworklifetime.

5.3. Remainingenergy

Figs.14–17presentthetotalremainingenergyofthenetworkwithrespecttothenumberofroundsofLEACHcompared withScenarios0through3respectivelywhenappliedasbasicorhybridwithc=5and10.

Thefigures clearlydemonstratethatthereis anextremesaveofenergyforour solutionoverLEACH.Notethat almost all graphsofLEACH andthebasicsolution havenearzero energyatthelast 3000to 5000 rounds,whichconfirmswhat we havediscussedearlieraboutthenumberofalivesensors graphs.Theremaining energygraphofthebasicsolutionof Scenario1isthesteepestwhencomparedtoallotherscenariosandhasabout15,000roundswithlessthan3%energy.This isbecause,asexplainedabove,Scenario1hasspecificareasinwhichnoMEsarepresentedforalongtime,andhence,the energydistributionofsendingsensorsisbiasedandsomenodesdiefasterthanothers.

5.4. Networkcoverage

Figs.18–21comparerandomlychosensnapshotsofthenetworkforbasicandhybridalgorithms ofScenarios0through 3respectively.Eachfigurehastwo sub-figures(a) and(b),suchthat fora chosennumberofrounds,sub-figure(a)shows thesnapshotofthenetworkforthebasicversionofthesolutionwhilesub-figure(b)representsthesnapshotofthehybrid solution. When comparing the sub-figures of each figure, it is clearlyshown that if there are some zones in the basic solution that are not covered duringsome rounds (shaded in gray), then they are guaranteed to be covered duringthe hybridsolution.Thezonesinblackarezonesthatare notcoveredby thedistributionofthesensorsintheWSNsettings.

(13)

(14)

ARTICLE IN PRESS

Fig. 14. Remaining energy versus rounds for LEACH, basic and hybrid solution of Scenario 0.

Fig. 15. Remaining energy sensors versus rounds for LEACH, basic and hybrid solution of Scenario 1.

(15)

(16)

ARTICLE IN PRESS

Fig. 18. Snapshots of basic and hybrid solution of Scenario 0 for rounds 2501 to 2505.

Hence,thesezonesaresaid tobeinactiveandnotconsidered inourexperimentalobservations.Eachother activezoneis markedwithaninteger.Thisintegerrepresentsthenumberofroundsatwhichatleastonesensorthatbelongstothiszone hassentdataduringthesnapshot.

(17)

Algorithm 1 Solution Algorithm.

6. Conclusionsandfuturework

Thispaperhasproposeda dataaggregationalgorithm forWSNsusingMobile Elements(MEs) intheIoT environment.

Theproposed algorithmisa cluster-basedaggregationalgorithm,which assignstherole ofClusterHead(CH)toarbitrary MEsthatpassthroughtheWSN.Thegoalofdoingsoismainlytosavetheenergyofsensorsand,therefore,toextendthe networklifetime,whileeffectivelyemployingtheexistingIoTtechnology.

Theproposedalgorithmhasbeendevelopedintotwophases,namely,thebasicalgorithmandthehybridalgorithm.The basiccaseassumestheavailability ofMEandsoalways assignstherole ofCHstotheMEs. Whereasthe hybridsolution realistically considers the caseofthe unavailability ofMEs. In thiscase, theCH role isswitched back to be assigned to regularsensorstoensurethat dataisnotlost.Thehybridsolutionalsodividesthenetworkintozones,suchthatifazone

(18)

ARTICLE IN PRESS

is notcovered by MEsfor adeﬁnedtime period,then it isconsidered asaseparate cluster andelectsone ofits regular sensorstobeaCH.

Simulationresultsshow that ourproposedsolution iseffectiveintermsofextendingthenetwork lifetime,andsaving theenergyofthesensors.ResultsalsopresentthatthealgorithmreactseﬃcientlytounavailabilityofMEsandguarantees thatdataisnotlost.Hence,ensuringthequalityofthedataaggregationofthenetwork.

Asforfuturework,weintendtoconsiderthebehavioroftheexternalmobileelementsthat passthroughthenetwork.

Ouraimistointroduceatrustlayerinthesysteminordertoselectthemostreliablemobileelementswhoarequaliﬁed enoughtotransmitthedata.Webelievethatthisselectionofthetrustedsensorsincreasesthetransmissionratesincethe datalossisexpectedtobereduced.

References

[1] Heinzelman WR , Chandrakasan A , Balakrishnan H . Energy-eﬃcient communication protocol for wireless microsensor networks. In: Proceedings of the 33rd Hawaii international conference on system sciences; 20 0 0. p. 3005–14 .

[2] Thein M , Thein T . An energy eﬃcient cluster-head selection for wireless sensor networks. In: Proceedings of international conference on intelligent systems, modelling and simulation, Liverpool, England; 2010. p. 287–91 .

[3] Xiangning F , Yulin S . Improvement on LEACH protocol of wireless sensor network. In: Proceedings of international conference on sensor technologies and applications, Valencia, Spain; 2007. p. 260–4 .

[4] Abdulsalam HM , Kamel LK . W-LEACH: Weighted low energy adaptive clustering hierarchy aggregation algorithm for data streams in wireless sensor networks. In: Proceedings of workshops of the 10th IEEE international conference on data mining (ICDM 2010), Sydney, Australia; 2010. p. 1–8 . [5] Abdulsalam HM , Kamel LK . Data aggregation for evolving wireless sensor networks using W-LEACH. In: Proceedings of the second Kuwait conference

on e-services and e-systems. Kuwait City, Kuwait; 2011. p. 1–5 .

[6] Abdulsalam HM , Ali BA . W-LEACH Based dynamic adaptive data aggregation algorithm for wireless sensor networks. Int J Distrib Sens Netw 2013 . [7] Pradhan J , Sarje A . A fault tolerant approach for data aggregation in wireless sensor networks. In: Proceedings of the international conference on

advances in computing, communications and informatics. Chennai, India; 2012. p. 382–6 .

[8] Jurdak R , Nafaa A , Barbirato A . Large scale environmental monitoring through integration of sensor and mesh networks. Sensors 2008;8(11):7493–517 . [9] Banerjee T , Xie B , Jun JH , Agrawal DP . Increasing lifetime of wireless sensor networks using controllable mobile cluster heads. Wireless Commun

Mobile Comput 2010;10(3):313–36 .

[10] Ma M , Yang Y . Clustering and load balancing in hybrid sensor networks with mobile cluster heads. In: Proceedings of the 3rd international conference on quality of service in heterogeneous wired/wireless networks. ACM; 2006 .

[11] Kim DS , Chung YJ . Self-organization routing protocol supporting mobile nodes for wireless sensor network. In: Computer and computational sciences, 2006. IMSCCS ’06. ﬁrst international multi-symposiums on, 2; 2006. p. 622–6 .

[12] Kumar GS , Vinu PMV , Jacob KP . Mobility metric based leach-mobile protocol. In: 16th International conference on advanced computing and communications; 2008. p. 248–53 .

[13] Ahmed A , Qazi S . Cluster head selection algorithm for mobile wireless sensor networks. In: International conference on open source systems and technologies; 2013. p. 120–5 .

[14] Di Francesco M , Das SK , Anastasi G . Data collection in wireless sensor networks with mobile elements: a survey. ACM Trans Sens Netw 2011;8(1):1–31 . [15] Ashton. K. 2009. “Internet of things. RFID J.”, http://www.rﬁdjournal.com/articles/view?4986 .

[16] Van Kranenburg R , Anzelmo E , Bassi A , Caprio D , Dodson S , Ratto M . The internet of things. A critique of ambient technology and the all-seeing network of RFID, Network Notebooks 2011;2 .

[17] Gubbi J , Buyya R , Marusic S , Palaniswami M . Internet of things (iot): a vision, architectural elements, and future directions. Future Generation Comput Syst 2013;29(7):1645–60 .

[18] Da Xu L , He W , Li S . Internet of things in industries: a survey. Ind Inf IEEE Trans 2014;10(4):2233–43 . [19] Atzori L , Iera A , Morabito G . The internet of things: a survey. Comput Netw 2010;54(15):2787–805 .

[20] Miorandi D , Sicari S , Pellegrini DF , Chlamtac I . Internet of things: vision, applications and research challenges. Ad Hoc Netw 2012;10(7):1497–516 . [21] Baronti P , Pillai P , Chook V , Chessa S , Gotta A , Hu YF . Wireless sensor networks: a survey on the state of the art and the 802.15. 4 and zigbee standards.

Comput Commun 2007;30(7):1655–95 .

[22] 2016. “AutoLog Data Sheet”, http://www.ff-automation.com/download/Documents/English/AutoLog _ DATASHEETS/FFAutomation _ DATASHEET _ AutoLog _ WiFi _ Sensor.pdf .

[23] DeCuir J . Introducing bluetooth smart: part 1: a look at both classic and new technologies. IEEE Consum Electron Mag 2014;3(1):12–18 .

[24] Siekkinen M , Hiienkari M , Nurminen JK , Nieminen J . How low energy is bluetooth low energy? comparative measurements with zigbee/802.15.4. In:

Wireless communications and networking conference workshops (WCNCW), 2012 IEEE; 2012. p. 232–7 .

[25] Yassein MB , Al-zou’bi A , Khamayseh Y , Mardini W . Improvement on leach protocol of wireless sensor network (vleach). Int J Digital Content Technol Appl 2009;3(2):132–6 .

(19)

Hanady M. Abdulsalam is an Assistant Professor in the Department of Information Science, Kuwait University, Kuwait. She received her B.Sc and M.Sc degrees in Computer Engineering from Kuwait University, and her Ph.D from the School of Computing, Queen’s University, Canada. Her research interests are in the areas of data management in WSN, internet of things, trust schemas and software engineering.

Bader A. Ali is an Assistant Professor in the Department of Information Science, Kuwait University, Kuwait. He received his B.Sc in computer engineering from UMC, Columbia, M.Sc in computer science from the USC, California, and his Ph.D in computer science from the School of Computer Science, McGill University. His research interests include large scale distributed systems, social networks and WSN.

Eman AlRoumi is an Teacher Assistant in the Department of Information Science, Kuwait University, Kuwait. She received her B.Sc and M.Sc degrees in Computer Engineering from Kuwait University. Her research interests include wireless sensor networks and internet of things.