Mobile charging information management for smart grid networks

(1)

Mobile

charging

information

management

for

smart

grid

networks

Shun-Neng

Yang

a,b,∗

_{, Hsiao-Wei}

_Wang

b

_{, Chai-Hien}

_Gan

a

_{, Yi-Bing}

_Lin

b

a_Information_and_{Communications}_Research_{Laboratories,}_ITRI,_Chutung,_Hsinchu,_31040,_Taiwan,_ROC b_Department_of_Computer_Science,_National_Chiao_Tung_University,_Hsinchu,_30010,_Taiwan,_ROC

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Available online 16 January 2013

Keywords: Smartgridnetwork Mobilechargingstation Electricvehicle

Mobilecharginginformationmanagement system

a

b

s

t

r

a

c

t

Withtoday’selectricchargingtechnology,chargingtimeofanelectricvehicle(EV)ismuchlongerthan thatforagasolinevehicle,andthereforethequeueingeffectatanEVchargingstation(CS)maybeserious. Thatis,whenanEVarrivesatanoverloadedCS,itislikelythattheEVwillwaitforalongtimebeforeit ischarged.ThispaperinvestigatesthewaitingproblemforEVcharging.WeproposeaMobileCS(MCS) managementsystemtodynamicallydistributechargingpolesupportthatreducesthewaitingtimesof EVsincurredinaﬁxedCS.AMobileChargingInformationManagementSystem(MC-IMS)ispresentedto describetheexecutionﬂowoftheMCSservice.Simulationexperimentsareconductedtoinvestigatethe waitingtimeperformancefortheproposedmechanism.OurstudyindicatesthattheMCS-basedMC-IMS provideseffectiveEVchargingwithshortwaitingtimes.

1. Introduction

Comparedwithgasolinecars,ElectricVehicles(EVs)provide

energy-efﬁcienttransportwithcarbonemissionreduction(Kim,

2003; Lee, 2007; Schoﬁeld, Yap, & Bingham, 2005; U.S. Fuel Economy,2011).AnimportantissueaboutEVchargingistodeploy

anefﬁcient smartgrid networkthat can effectivelyand

conve-nientlychargetheEVs.ThisGrid-to-Vehicle(G2V)issuehasbeen

intensivelyinvestigatedintheliterature(Verma,Singh,&Shahani,

2011).OtherEVcharging/dischargingissuesforG2V-basedsmart

gridnetwork havebeen alsostudied, includingVehicle-to-Grid

(V2G)andVehicle-to-Vehicle(V2V)scenarios.AnEVisgenerally

equippedwithenergystorageof24–56kWh(NissanUSA,2012;

TeslaMotors,2011),whichisabout2–5timesmorethanthe

aver-agedailyenergy consumption ofa householdin Taiwan(Miao,

2008).V2GtechnologyutilizessuchhighenergystorageofEVto

dischargeelectricitytothepowergrid(Su,Eichi,Zeng,&Chow,

2008).IntheV2Vtechnology,anEVcansoakupthepowerfromthe

batteryofaspeciﬁcEVthatisequippedwithplentyenergystorage

(Li,Sahinoglu,Tao,&Teo,2010;Sahinoglu,Tao,&Teo,2010).The

V2GandV2Vtechnologiesarecurrentlyunderfeasibilitystudyand

trialoperations(Baker,2011;Lietal.,2010;NUVVECorporation,

2012;Sahinogluetal.,2010;Suetal.,2008).Thispaperfocuseson

G2VarchitecturethatalsoaccommodatestheV2Vtechnology.

∗ Correspondingauthorat:Rm.505,Bldg.51,No.195,Sec.4,ChungHsingRoad, Chutung,Hsinchu,31040,Taiwan,ROC.Tel.:+88635912286;fax:+88635820263.

E-mailaddresses:[email protected],[email protected]

(S.-N.Yang),[email protected](H.-W.Wang),[email protected](C.-H.Gan),

[email protected](Y.-B.Lin).

Inasmartgridnetwork,severalFixedChargingStations(FCSs)

aredeployedingeographicareasuchasahighway.Thecharging

time of an EV typically exceeds 30min with the present

fast-chargingtechnologies(Nor,1993;Winkleretal.,2009).IfanEV

arrivesatanoverloadedFCS,it willwait foralongtime before

itischarged.ThispaperintegratesG2VandV2Vtechnologiesto

reduce theEV waiting timefor charging.The ideais todeploy

MobileChargingStations(MCSs)inthesmartgridnetwork,which

aredynamicallydistributedtorelievetheworkloadofbusyFCSs.

AnMCSisaspeciﬁcEVequippedwithseveralfastChargingPoles

(CPs).WhentoomanyEVsarriveatanFCS(speciﬁcallythe

num-berofEVsintheFCSexceedsatolerablethreshold),anMCSwillbe

dispatchedtosupporttheFCSforrelievingthechargingworkload.

ThepowersourceofanMCSmaycomefromtheconnectiontothe

powergridattheFCSortheMCScanbeequippedwithlimited

energystorage.Thispaperstudiesthescenariooftheformercase.

Inthelattercase,theMCSneedstoberechargedfromtimetotime.

InformationManagement System(IMS) hasbeen utilized to

supportmanylarge-scaleschedulingandparallelprocessing

appli-cationssuchasaviation(Abdi&Sharma,2007)andcryptography

(Ogiela&Ogiela,2012).Inthispaper,wetaketheadvantageofthe

IMSapproachfordispatchingtheMCSs.Speciﬁcally,wedevelopa

MobileChargingInformationManagementSystem(MC-IMS)that

dynamicallydispatchestheMCSstorelievetheworkloadofFCSs.

Inthissolution,anEVisconnectedtoanFCSforG2Vcharging,and

isconnectedtoanMCSforV2Vcharging.Ourapproachdistributes

MCSsbasedontheworkloadinformation(i.e.,thenumberofEVsin

theFCS)ofFCSs.Weconductsimulationexperimentstocompare

theperformancebetweentraditionalFCSsystemandtheMC-IMS

intermsoftheaveragewaitingtimeofEVs.Inthispaper,thesmart

gridnetworkiscalledanFCSnetworkifitisonlydeployedwith

(2)

246 S.-N.Yangetal./InternationalJournalofInformationManagement33 (2013) 245–251

Fig.1. TheMC-IMS-basedMCSNetworkArchitecture.

FCSs(andthenetworkonlysupportsG2Vcharging).Ontheother

hand,ifMCSsarealsodeployedbesidesFCSstobedispatchedby

theMC-IMS,thenthesmartgridnetworkiscalledanMCSnetwork

(wherebothG2VandV2Vchargingaresupported).

2. MC-IMS-basedsmartgridnetwork

ThissectiondescribesthewaitingproblemofEVsatFCSs.Then

wedescribehowtheMC-IMSdispatchesMCSsbasedonthe

work-loadinformationofFCSs.

Withoutlossofgenerality,weconsiderthe

Taiwan-highway-likemodel(TaiwanAreaNationalFreewayBureauofMOTC,2012)

toshowtheMC-IMSarchitecture.AsillustratedinFig.1,thelength

ofthehighwayis520kmwith12intermediatehighwayrestareas.

WeassumethattheFCSsarelocatedatthehighwayrestareas(one

FCSperrestarea).ThedistancebetweentwoFCSsis40km.

Inasmartgridnetwork,anEVisinstalledanon-boardunit(OBU,

aGPS-basednavigator).Therefore,fromadigitalmapandGPS,the

EVisawareofthelocationsoftheFCSs.InanFCSnetwork,when

anEVtravelsonthehighway,itsOBUcanselectthenextFCSfor

chargingbasedonthelocalinformation(i.e.,thecurrentposition

oftheEV,theremainingelectricity,andthedistancestotheFCSs)

(Imai,Ashida,Zhang,&Minami,2008).Fromthelocalinformation,

theOBUcompilesalistofreachableFCSs,andthenselectsthenext

FCSforcharging.WeassumethateachEVselectsthefarthestFCS

forcharging.

Since each EV selects thenext FCS without consideringthe

queueingstatusofFCS,theEVsmayselectbusyFCSs.Toresolve

thisissue,theMCSsareintroducedintotheFCSnetworkandare

dispatchedbytheMC-IMS(seeFig.1(1))tosupporthot-spotFCS

locations.AnFCS(e.g.,Fig.1(2))requestsMCSsupportfromthe

MC-IMSwhenthenumberofEVsinthisFCSislargerthanahigh

threshold.ThentheMC-IMSwilldispatchanMCStoprovideextra

chargingcapacity totheFCSlocation.On theotherhand,ifthe

workloadofanFCSwithMCSsisunderalowthreshold,thenthe

MCSsatthisFCSlocationcanbedispatchedtosupportotherhot

spots.

Theparametersusedinthismechanismaredescribedasfollows.

SupposethatthereareFFCSs,andeachFCShasPfCPs.ThereareM

MCSs,eachofthemhasPmCPs.TheithFCSisdenotedasf(i),where

1≤i≤F.Inf(i),severalparametersaremaintained:

• q:thequeuelength(theworkload)orthenumberofEVsinthe

FCS

• f:aﬂagindicatingthestatusoftheworkload,wheref=0

repre-sentsthelight-load,f=1representsthemedium-load,andf=2

representstheheavy-load

• Pn:thenumberofCPsatanFCS’slocation,wheref(i)·Pn=Pf+jPm

meansthattherearejMCSsatf(i)’slocation

• LR:theMCS-requestinglock,whereLR=0(LR=1)representsthat

theFCScan(cannot)requestMCSsupport

TodeterminewhetheranFCSwillrequestanMCSordonatean

MCS,thecriterionwithaweightedfactor˛>1isconsideredto

indi-catetheworkloadofanFCS.AnFCSisheavy-loadiff(i)·q≥˛f(i)·Pn.

Ontheotherhand,theFCSislight-loadiff(i)·Pn–f(i)·q≥Pm.A

light-loadFCScandonateMCSsiff(i)·Pn>Pf(i.e.,thereisatleastoneMCS

atthisFCSlocation).TheparametersmaintainedintheMC-IMS

include:

• QD:thedonorlistorthesetoflight-loadFCSsthatcandonateone

ormoreMCSs

• QR:therequesterlistorthesetofheavy-loadFCSsthatrequire

extraCPs

Basedontheaboveparameters,wedeveloptheproceduresfor

theMCS-Dispatchingmechanismtobedescribedinthenext

sec-tion.

3. TheproceduresfortheMCS-Dispatchingmechanism

Theproceduresfor theMCS-Dispatchingmechanism include

EV-Arrival, Request-MCS, Dispatch-MCS, Donate-MCS,

Cancel-Donate-MCS, Charge-Complete, Cancel-Request-MCS, and

MCS-Arrival.Thedetailsareelaboratedinthesubsequentsubsections.

3.1. EV-Arrivalprocedure

WhenanEVarrivesatf(i)forcharging,theEV-Arrivalprocedure

isexecuted.ThemessageﬂowisshowninFig.2andisdescribedas

follows:

StepA.1.AnEVarrivesatf(i),andf(i)·qisincrementedby1.

StepA.2.(Checkiftheworkloadoff(i)changesfrommedium

toheavy).Iff(i)·q≥˛f(i)·Pnandf(i)·f=1,thenf(i)requestsextraCPs

fromtheMC-IMSwiththefollowingactions:

StepA.2.1.f(i)·fissetto2.

StepA.2.2.Iff(i)·LR=0(i.e.,therenoMCSdispatchedtof(i)and

driv-ingtof(i)’slocation),thenf(i)invokestheRequest-MCSprocedure

(seeSection3.2)andexits.

StepA.3.(Checkiftheworkloadoff(i) changesfromlightto

medium).Elseiff(i)·Pn–f(i)·q<Pmandf(i)·f=0,thenthefollowing

(3)

Fig.2.ThemessageﬂowoftheEV-Arrivalprocedure.

StepA.3.1.f(i)·fissetto1.

StepA.3.2.Iff(i)·Pn>Pf(i.e.,thereareoneormoreMCSsatf(i)’s

location),thenf(i)invokestheCancel-Donate-MCSprocedure(see

Section3.5)andexits.

Notethatiff(i)·q≤f(i)·Pn afterStepA.1isexecuted,thenthe

EVisservedbyaCPimmediately.Otherwise,theEVisqueuedin

f(i).Whentheworkloadoff(i) changesfromlighttomedium,if

f(i)·Pn=PfatStepA.3,thereisnoMCSatf(i)’slocation,andf(i)does

notdonateanyMCStotheMC-IMS.

IfneitherStepA.2norStepA.3isexecuted,itmeansthatthe

workloadoff(i)doesnotchangewhentheEVarrives,andf(i)needs

nottointeractwiththeMC-IMS.

3.2. Request-MCSprocedure

In the Request-MCS procedure, f(i) requests the MC-IMS to

provideMCSsupport.TheMC-IMSselectsalight-loadFCSfrom

QDtodonateextraMCSs.ThemessageﬂowisshowninFig.3and

isdescribedasfollows:

StepB.1.f(i)sendstheRequest-MCSmessagetotheMC-IMSfor

MCSsupport.

StepB.2.UponreceiptoftheRequest-MCSmessage,theMC-IMS

checksthestatusofQD.

StepB.3.IfQD=(i.e.,noFCScandonateanyMCStof(i)),then

theMC-IMStakesthefollowingactions:

StepB.3.1.QRissettoQR∪{i}.

StepB.3.2.TheMC-IMSsendstheRequest-MCS-Responsemessage

withtheresult“Reject”tof(i),andthisprocedureexits.

StepB.4.Else(i.e.,QD /= andthereisatleastanFCSthatcan

donateanMCS)theMC-IMSselectskinQDsuchthatthedistance

off(k)isclosesttof(i).

StepB.4.1.TheMC-IMSinvokestheDispatch-MCSprocedure(see

Section3.3)thatinstructsf(k)tosendanMCStof(i).

StepB.4.2.CheckthereturnedresultcoftheDispatch-MCS

pro-cedure(c≥0representsthenumberofMCSsremainedatf(k)’s

locationafterdispatchinganMCStof(i),andc<0representsthat

nofreeMCSisatf(k)’slocation).

StepB.4.3.Ifc≤0(i.e.,f(k)hasmediumorheavyworkload),then

QDissettoQD–{k}.

Fig.3.ThemessageﬂowoftheRequest-MCSprocedure.

StepB.4.4.Ifc≥0(i.e.,anMCShasbeendispatchedtosupport

f(i)),thentheMC-IMSsendstheRequest-MCS-Responsemessage

withtheresult“Accept”tof(i).Uponreceiptofthe

Request-MCS-Responsemessagewiththeresult“Accept”,f(i)setsf(i)·LRto1.

StepB.4.5.Else(i.e.,theMC-IMSshouldselectanotherFCSfrom

QDtosupportf(i))gotoStepB.3.

NotethatalthoughQDindicatesthatanFCSf(k)candonatean

MCSatStepB.4,f(k)maycancelitsdonationbeforeStepB.4.1is

completed.Ifso,theMC-IMSmustcheckQDtoﬁndanotherMCS

donor.Therefore,StepB.4.5mayloopbacktoStepB.3.AtStepB.4.3

andB.4.4,whenc=0,itmeansthatf(k)becomesmedium-loadafter

ithasdispatchedanMCS.

3.3. Dispatch-MCSprocedure

In the Dispatch-MCS procedure, the MC-IMS dispatches an

MCS from f(k) to f(i). In this procedure, we can use the

switching technique to switch the EVs charged by the CPs

of certain MCSs to the CPs of the FCS or other MCSs, and

[f(k)·Pn−max(f(k)·q,Pf)]/Pm

isthenumberofMCSsthatf(k)

candonate(if

≥1).Themessage

ﬂowisshowninFig.4andisdescribedasfollows:

StepC.1.TheMC-IMSsendstheRequest-MCSmessagetof(k).

Thismessagerequestsf(k)todispatchanMCStosupportf(i).

StepC.2.UponreceiptoftheRequest-MCSmessage,f(k)checks

itsworkload.

StepC.3.If

[f(k)·Pn−max(f(k)·q,Pf]/Pm

≥1 (i.e.,f(k)can

donateatleastoneMCS),thenf(k)dispatchesanMCStof(i)and

f(k)·Pnissettof(k)·Pn–Pm.

StepC.4.If

=1(i.e.,thestatus

off(k)willbecomemediumafteritdonatestheMCS),thenf(k)·fis

setto1.

StepC.5.f(k)sendstheRequest-MCS-Responsemessagewith

thestatuscodec=

tothe

MC-IMS.

AtStepC.3,f(k)_·Pn isdecreasedbytheamountPm.Therefore,

atStepC.5,c<0representsthatnofreeMCSisatf(k)’slocation,

(4)

Fig.4.ThemessageﬂowoftheDispatch-MCSprocedure.

representsthatmorethanonefreeMCSsareatf(k)’slocation,and

f(k)hasdispatchedanMCSforc≥0.

3.4. Donate-MCSprocedure

In theDonate-MCS procedure, f(i) informsthe MC-IMS that

f(i)candonateMCSs.ThemessageﬂowisshowninFig.5andis

describedasfollows:

StepD.1.f(i)sendstheDonate-MCSmessagetotheMC-IMS.

StepD.2.UponreceiptoftheDonate-MCSmessage,theMC-IMS

checksthestatusofQR.

StepD.3.IfQR=(i.e.,noFCSneedsextraCPs),thenQDissetto

QD∪{i}.

StepD.4.Else(i.e.,QR /=andthereisanFCSthatneedsextra

CPs)theMC-IMSselectskfromQR.

StepD.4.1.TheMC-IMSinvokestheDispatch-MCSprocedure(see

Section3.3)thatinstructsf(i)tosendanMCStof(k).

StepD.4.2.TheDispatch-MCSprocedurereturnstheresultctothe

MC-IMS.

StepD.4.3.Ifc≥0(i.e.,f(i)hasdispatchedanMCStosupportf(k)),

thentheMC-IMSsetsQR toQR–{k}andsendsthe

Dispatching-Notiﬁcationmessage to notify that an MCS is dispatched and

drivingtof(k).UponreceiptoftheDispatching-Notiﬁcation

mes-sage,f(k)setsf(k)·LRto1.

StepD.5.TheMC-IMSsendstheDonate-MCS-Responsemessage

tof(i).

Fig.5. ThemessageﬂowoftheDonate-MCSprocedure.

Fig.6.ThemessageﬂowoftheCancel-Donate-MCSprocedure.

AfterStepD.4.2,ifc<0(i.e.,f(i)cannotdonateanyMCSto

sup-portf(k)),theMC-IMStakesnoaction.

3.5. Cancel-Donate-MCSprocedure

IntheCancel-Donate-MCSprocedure,f(i)informstheMC-IMS

tocancelthepreviousdonation.Whenthisprocedureisexecuted,

itmeansthataftertheFCSdecidedtodonateanMCS,itsworkload

hasbeenincreased,andtheFCScannolongerkeepitspromise(and

thereforehastocancelthedonation).Themessageﬂowisshown

inFig.6andisdescribedasfollows:

StepE.1.f(i)sendstheCancel-Donate-MCSmessagetoinform

theMC-IMSthatf(i)willnotreleaseanyMCS.

StepE.2.UponreceiptoftheCancel-Donate-MCSmessage,the

MC-IMS sets QD to QD–{i} and sendsthe

Cancel-Donate-MCS-Responsemessagetoinformf(i)thatthedonationiscanceled.

3.6. Charge-Completeprocedure

When an EV completes its charging from f(i), the

Charge-Completeprocedureis executed.Themessageﬂowis shownin

Fig.7andisdescribedasfollows:

StepF.1. AnEVcompletesitschargingfromf(i), andf(i)·qis

decrementedby1.

StepF.2.(Checkiftheworkloadoff(i)changesfrommediumto

light).Iff(i)·Pn–f(i)·q≥Pmandf(i)·f=1,thenthefollowingsubsteps

areexecuted:

StepF.2.1.f(i)·fissetto0.

(5)

Fig.8.ThemessageﬂowoftheCancel-Request-MCSprocedure.

StepF.2.2.Iff(i)·Pn>Pf(i.e.,thereareoneormorefreeMCSsatf(i)’s

location),thenf(i)invokestheDonate-MCSprocedure(seeSection

3.4)andexits.

StepF.3.(Checkiftheworkloadoff(i)changesfromheavyto

medium).Elseiff(i)_·q<_˛f(i)·Pn andf(i)·f=2,thenf(i)informsthe

MC-IMSthatitnolongerrequestsforMCSsupportwiththe

fol-lowingactions:

StepF.3.1.f(i)·fissetto1.

StepF.3.2. f(i) invokestheCancel-Request-MCS procedure(see

Notethatwhentheworkloadoff(i)changesfrommediumto

light,iff(i)·Pn=PfatStepF.2,thereisnoMCSatf(i)’slocation,and

f(i)needsnottointeractwiththeMC-IMS.

IfneitherStepF.2norStepF.3isexecuted,itmeansthatthe

workload off(i) doesnot change.Inthis case, f(i) needsnotto

interactwiththeMC-IMS.

3.7. Cancel-Request-MCSprocedure

IntheCancel-Request-MCSprocedure,f(i)informstheMC-IMS

tocancelthepreviousRequest-MCScommandsentfromf(i).When

this procedureis executed,it meansthatafter theFCSdecided

torequestanMCS,itsworkloadhasbeendecreased,andtheFCS

nolongerrequestsforMCSsupport(andthereforehastocancel

thepreviousrequest).ThemessageﬂowisshowninFig.8andis

describedasfollows:

StepG.1.f(i)sendstheCancel-Request-MCSmessagetoinform

theMC-IMSthatf(i)willnotrequestextraMCSs.

Step G.2.Upon receipt of theCancel-Request-MCS message,

theMC-IMSsetsQRtoQR–{i}andsendsthe

Cancel-Request-MCS-Responsemessagetoinformf(i) thattherequest issuccessfully

canceled.

3.8. MCS-Arrivalprocedure

WhenanMCSarrivesatf(i)toprovideextraCPs,theMCS-Arrival

procedureisexecuted.ThemessageﬂowisshowninFig.9andis

describedasfollows:

StepH.1.AnMCSarrivesatf(i)toprovideextraCPs.f(i)·Pnisset

tof(i)·Pn+Pm,andf(i)·LRissetto0.

Step H.2. (Check if the workload of f(i) is still heavy). If

f(i)_·q_≥_˛f(i)·Pn,thenf(i)invokestheRequest-MCSprocedure(see

StepH.3.(Checkiftheworkloadoff(i)changesfrommedium

tolight).Elseiff(i)·Pn–f(i)·q≥Pmandf(i)·f=1,thenf(i)candonate

MCSswiththefollowingactions:

StepH.3.1.f(i)·fissetto0.

StepH.3.2.f(i)invokestheDonate-MCSprocedure(seeSection3.4)

andexits.

Fig.9.ThemessageﬂowoftheMCS-Arrivalprocedure.

StepH.4.(Checkiftheworkloadoff(i)changesfromheavyto

medium)Elseiff(i)·q<˛f(i)·Pnandf(i)·f=2,thenf(i)·fissetto1.

IfnoneofthestepsinStepH.2,StepH.3,andStepH.4isexecuted,

itmeansthattheworkloadoff(i)isstillmediumorstilllight.Inthis

case,f(i)needsnottointeractwiththeMC-IMS.

4. Performanceevaluation

Wedevelopanevent-drivensimulationmodel(Gan&Lin,2007)

tocomputetheoutputmeasuresfortheFCSandtheMCSnetworks.

Basedonthesimulationexperiments,thissectioncomparesthe

performanceofthesetwonetworks.Inoursimulationexperiments,

thespeedofanEVisuniformlydistributedbetween60km/hand

100km/h,whicharethelowerandtheupperspeedlimitsofTNE1,

respectively.TheinitialpowerofanEVisuniformlydistributed

between25% and100% (thevalue 25%is theminimalrequired

powerforanEVtodrivetothenearestFCS).Afully-chargedEVcan

lastfor160km(e.g.,whenanEVtravelsforatriplongerthan160km

inthehighway,itmustberechargedattheFCSs),andthetimefor

anEVchargingfrom0%to100%is30min.EachEVischargedto

100%atanFCS,andthechargingtimeislinearlyproportionalto

theamountofchargedenergy.Wesimulate1,000,000EVarrivals.

FortheFCSworkloadindication,theweightedfactor˛issetto1.5.

TheinputparametersinthesimulationexperimentsaretheEV

arrivalrate(),thenumberofFCSs(F),thenumberofCPsateach

FCS(Pf),thenumberofMCSs(M),andthenumberofCPsoneach

MCS(Pm).

Themajoroutputmeasureinthispaperistheaveragewaiting

timeWforanEVbeforeitischargedatanFCSlocation.

Withoutlossofgenerality,therearetwotypesofEVtrafﬁc

pat-terns:Intheﬁrstpattern,theEVsdrivefrompointAtopointCas

illustratedinFig.1.Inthesecondpattern,theEVsdrivefrompoint

BtopointD.Thelengthofeachpathis340km.Notethatthepaths

(A,C)and(B,D)haveanoverlappingsegment(B,C)thatcoversfour

FCSs(i.e.,fromFCS5toFCS8).TheEVsarrivalsareaPoissonprocess

withtheraterangingfrom1.02to1.06,whichisabout8.5–8.8%

oftheaveragevehicletrafﬁcofSijhihatTaiwanNational

Express-way1(TNE1)insouthwarddirection(i.e.,foreveryminute,there

are12cararrivals,andamongthem,1.02–1.06areEVs)(Ministry

ofTransportation&Communications,2011).EachEVselectsthe

farthestFCSforcharging.WhenanEVentersanFCSandthenis

servedbyaCP,wecollectitswaitingtimeattheFCSlocation(i.e.,

thetimethattheEVwaitsforcharging).

Weassumethatthereare12FCSs,andeachFCShas20CPsin

(6)

Fig.10.TheCPdistributionamongtheFCSlocations(=1.04;FCSnetwork:F=12, Pf=20,M=Pm=0;MCSnetwork:F=12,Pf=17,M×Pm=36).

MCSsandeachMCSisequippedwithPmCPs.ThespeedofanMCS

issetto90km/h,whichistheupperspeedlimitoflargetruckon

TNE1.Tomakeafaircomparison,thetotalnumberofCPsinthe

MCSnetworkisalsosetto240inthefollowingsimulations(i.e.,

F×Pf+M×Pm=240,whereFisﬁxedto12).

Fig.10comparestheCPdistributionbetweentheFCSnetwork

andtheMCSnetwork.NotethatintheFCSnetwork,everyFCShas

20CPs(seetheline).IntheMCSnetwork,weallocate15%oftotal

CPsto2,4,and6MCSsrespectively(i.e.,eachMCShas18,9,and

6CPs,respectively)andthearrivalrateisﬁxedto1.04.Sincethe

overlappingsegment(B,C)coversFCS5toFCS8,itismorelikelythat

thehotspotsoccurinthesefourFCSs.Fig.10showsthatFCS5to

FCS7receivemoreextraCPsthatwillmitigateworkloadsofthehot

spots.TheMCSnetworkcandistributeCPsﬂexiblysincethemobile

CPscanbedonatedtothehot-spotFCSsandeffectivelyreducesthe

queueingeffectintheseFCSs.

Fig.11comparestheaveragewaitingtimeWbetweentheFCS

networkandtheMCSnetwork,wherethetotalnumberofCPsis

ﬁxedto240.Inthissimulationexperiment,theEVarrivalrate

rangesbetween1.02and1.06.Weallocate15%oftotalCPsto2,4,

and6MCSs,respectively.Intuitively,Wincreasesasincreases.In

theFCSnetwork,ifissmall,Wincreasesslowlyasincreases.

When>1.05,Wincreasesfastasincreasessincethequeueing

effectsofthehotspotsareintensiﬁedwiththehighEVarrivalrates.

Ontheotherhand,theMCSnetworkcaneffectivelydistributethe

CPstothehotspots(i.e.,FCS5toFCS7)toreducethequeueingeffect

Fig.11. TheWperformance(FCSnetwork:F=12,Pf=20,M=Pm=0;MCSnetwork: F=12,Pf=17,M×Pm=36).

oftheseFCSs(seeFig.10).Therefore,theWvaluesintheMCS

net-workincreaseinsigniﬁcantlyforallvalues,andaremuchlower

thanthoseintheFCSnetwork.TheadvantageoftheMCSnetwork

overtheFCSnetworkbecomesverysigniﬁcantwhentrafﬁcloadis

heavy.

LetWFbetheWfortheFCSnetwork,andWMbetheWforthe

MCSnetwork.ThentheimprovementofWbytheMCSnetwork

overtheFCSnetworkcanbedeﬁnedas:

I=WF−WM

WF (1)

Fig.12(a)showsthewaitingtimeimprovementIforthevarious

percentagesofCPscarriedbytheMCSs,wherethetotalnumberof

CPsisﬁxedto240,andweconsiderthescenarioswhere5%,10%,

15%and20%ofCPsareequippedonMCSsintheexperiments.The

EVarrivalrate␭isﬁxedto1.04.Clearly,whenthepercentageof

mobileCPsisﬁxed,theimprovementIofthewaitingtimeincreases

withthenumberoftheMCSsM.

Fig.12(b)showsthewaitingtimeimprovementIasthefunction

ofMCSnumber.ComparedtotheFCSnetwork,theimprovement

ofWperformanceissigniﬁcantlyimprovedbytheMCSnetwork

with10%mobileCPsallocatedtotheMCSs;thatis,allocatinga

smallamountofCPstoMCSscanobtainsigniﬁcantlyperformance

improvement. Thus ifthebudget for buildingan MCSnetwork

islimited, andtheMCSnetwork operator wouldliketoreduce

(7)

ThispaperproposedtwotypesofsmartgridnetworksforanEV

charging:theFixedChargingStation(FCS)andtheMobile

Charg-ingStation(MCS)networks.TheFCSnetwork onlyutilizesﬁxed

chargingstations.TheMCSnetworkcombinesFCSswithmobile

chargingstations dispatchedby a Mobile Charging Information

ManagementSystem(MC-IMS).Thesimulationexperimentsare

conductedtoinvestigatethewaitingtimeperformanceforthese

smartgridnetworks.OurexperimentsindicatedthattheMCS

net-workhasbetterwaitingtimeperformancethantheFCSnetwork.

TheadvantageoftheMCSnetworkovertheFCSnetworkbecomes

verysigniﬁcantwhentheEVarrivalrateislarge.

Inthispaper,everyFCSlocationhasinﬁnitewaitingcapability.

Inthefuture,wewillextendourstudytoconsiderFCSlocations

wheretheparkingspacesarelimited.

Acknowledgements

ThispaperwassupportedinpartbytheITRI/NCTUJRCResearch

Project,theITRIAdvancedResearchProgramunderB301EA3300,

B301AR2R10, and B352BW1100, under C352BW1100, C301EA

3600, and C301AR2D10, ICL/ITRI, NSC 100-2221-E-009-070,

ChunghwaTelecom,IBM,ArcadyanTechnologyCorporation,Nokia

SiemensNetworks,Department of Industrial Technology(DoIT)

AcademicTechnologyDevelopmentProgram

100-EC-17-A-03-S1-193,theMoEATUplan,andtheTechnologyDevelopmentProgram

oftheMinistryofEconomicAffairs(MoEATDP),Taiwan.

References

Abdi,M.R.,&Sharma,S.(2007).Strategic/tacticalinformationmanagementofﬂight operationsinabnormalconditionsthroughNetworkControlCentre. Interna-tionalJournalofInformationManagement,27(2),119–138.

Baker, S. (2011). V2G and G2V: It’s about grid scale storage. Retrieved on 2012-05-20, from http://bpiconference.com/blog/wp-content/uploads/2011/ 10/Baker Scott.pdf

Gan,C.-H.,&Lin,Y.-B.(2007).Push-to-Talkserviceforintelligenttransportation systems.IEEETransactionsonIntelligentTransportSystems,8(3),391–399. Imai,K.,Ashida,T.,Zhang,Y.,&Minami,S.(2008).EVrangeextender:bettermileage

thanplug-inhybrid?InPaperpresentedattheIEEEvehiclepowerandpropulsion conferenceHeiLongjiang,China,(pp.1–3).

Kim,P.-S.(2003).Costmodelingofbatteryelectricvehicleandhybridelectricvehicle basedonmajorpartscost.InPaperpresentedattheﬁfthinternationalconference onpowerelectronicsanddrivesystems(pp.1295–1300).

Lee,M.-Y.(2007).ElectricVehicleismuchbetterinTaiwan.Retrievedon2012-05-20, fromhttp://sa.ylib.com/MagCont.aspx?Unit=featurearticles&id=1712 Li,Z.,Sahinoglu,Z.,Tao,Z.,&Teo,K.H.(2010).Electricvehiclesnetworkwithnomadic

portablechargingstations.InPaperpresentedattheIEEE72ndvehicular technol-ogyconferencefallOttawa,Canada,(pp.1–5).

Miao,Y.-Y.(2008).Windpowergenerationyearlysupportstheenergy consump-tionof230thousandhouseholdsinTaiwan.Retrievedon2012.05.20,from

http://e-info.org.tw/node/37206

MinistryofTransportationandCommunications(2011).Monthlystatisticsof trans-portationandcommunications.Retrievedon2012.05.20,from http://motc. gov.tw/uploaddowndoc?file=mebook/10004book.pdf&filedisplay=10004book. pdf&flag=doc

Nissan USA (2012). The new car: features and speciﬁcations. Retrieved on2012.05.20,from http://www.nissanusa.com/leaf-electric-car/index#/leaf-electric-car/specs-features/index

Nor,J.K.(1993).Artofchargingelectricvehiclebatteries.InPaperpresentedatthe WESCON/’93conferenceCalifornia,USA,(pp.521–525).

NUVVECorporation(2012).SMARTGRIDpoweredbyV2G.Retrievedon2012.05.20, fromhttp://nuvve.com/

Ogiela,M.R.,&Ogiela,L.(2012).DNA-likelinguisticsecretsharingforstrategic informationsystems.InternationalJournalofInformationManagement,32(2), 175–181.

Taiwan Area National Freeway Bureau of MOTC (2012). The ofﬁcial site of TaiwanareanationalfreewaybureauofMOTC.Retrievedon2012.05.20,from http://www.freeway.gov.tw/

TeslaMotors(2011).Increasingenergydensitymeansincreasingrange.Retrieved on2012.05.20,fromhttp://www.teslamotors.com/roadster/technology/battery U.S. Fuel Economy (2011). Electric vehicles. Retrieved on 2012.05.20, from

http://www.fueleconomy.gov/feg/evtech.shtml

Verma,A.K.,Singh,B.,&Shahani,D.T.(2011).Gridtovehicleandvehicletogrid energytransferusingsingle-phasebidirectionalACDCconverterand bidirec-tionalDC–DCconverter.InPaperpresentedattheinternationalconferenceon energy,automation,andsignalBhubaneswar,India,(pp.1–5).

Winkler,T.,Komarnicki,P.,Mueller,G.,Heideck,G.,Heuer,M.,&Styczynski,Z.A. (2009).ElectricvehiclechargingstationsinMagdeburg.InPaperpresentedat theIEEEvehiclepowerandpropulsionconference(pp.60–65).

Shun-NengYangreceivedtheB.S.andtheM.S.degrees inComputerScienceandInformationEngineeringfrom NationalChengKungUniversity(NCKU),Tainan,Taiwan, R.O.C.,in2004and2006,respectively.Heiscurrently workingtowardthePh.D.degreeatNCTU.Heisalso aResearcher intheInformationand Communications ResearchLabs,IndustrialTechnologyResearchInstitute (ICL/ITRI).Hiscurrentresearchinterestsincludesmart gridnetwork,wirelessandmobilecomputing,personal communicationsservices,andperformancemodeling.

Hsiao-Wei Wang received the B.S.C.S. degree from NationalChiaoTungUniversity(NCTU),Hsinchu,Taiwan, R.O.C.,in 2012. She is currently working towardthe M.S.C.S.degreeatNCTU.Hercurrentresearchinterests includedesignandanalysisofsmartgridnetworkand performancemodeling.

Chai-HienGan(M’05)receivedhisBSdegreein com-putersciencefromTamkangUniversityin1994,Taipei County,Taiwan,and bothhis M.S. andPh.D. degrees incomputerscienceandinformationengineeringfrom NationalTaiwanUniversity,Taipei,Taiwan,in1996and 2005,respectively.FromMarch2005toJuly2007,hewas aResearchAssistantProfessorinDepartmentofComputer Science,NationalChiaoTungUniversity,Taiwan.Since July2007,hehasbeenaResearcherintheInformationand CommunicationsResearch Labs,Industrial Technology ResearchInstitute(ICL/ITRI),Taiwan.Hiscurrentresearch interestsincludewirelessandmobilecomputing,personal communicationsservices,IPmultimediasubsystem,and wirelessInternet.

Yi-BingLinisVicePresidentandLifeChairprofessorof CollegeofComputerScience,NationalChiaoTung Uni-versity(NCTU),anda VisitingprofessorofKingSaud University.HeisalsowithInstituteofInformation Sci-enceandtheResearchCenterforInformationTechnology Innovation,AcademiaSinica,Nankang,Taipei, Taiwan, R.O.C. Lin is the authors of the books Wireless and MobileNetworkArchitecture(Wiley,2001),Wirelessand MobileAll-IPNetworks(JohnWiley,2005),and Charg-ingforMobileAll-IPTelecommunications(Wiley,2008). Linreceivednumerousresearchawardsincluding2005 NSCDistinguishedResearcherand2006AcademicAward ofMinistryofEducation.LinisanACMFellow,anAAAS Fellow,anIEEEFellowandanIETFellow.