• Tidak ada hasil yang ditemukan

Interference from cloud-to-ground and cloud flashes in wireless communication system.

N/A
N/A
Protected

Academic year: 2017

Membagikan "Interference from cloud-to-ground and cloud flashes in wireless communication system."

Copied!
10
0
0

Teks penuh

(1)

ContentslistsavailableatScienceDirect

Electric

Power

Systems

Research

jou rn al h om e p a g e :w w w . e l s e v i e r . c o m / l o c a t e / e p s r

Interference

from

cloud-to-ground

and

cloud

flashes

in

wireless

communication

system

Mohd

Riduan

Ahmad

a,b,∗

,

Mona

Riza

Mohd

Esa

a,c

,

Vernon

Cooray

a

,

Eryk

Dutkiewicz

d

aÅngströmLaboratory,DivisionforElectricity,DepartmentofEngineeringSciences,UppsalaUniversity,Box534,S-75121,Sweden bUniversitiTeknikalMalaysiaMelaka,HangTuahJaya,76100DurianTunggal,Malacca,Malaysia

cUniversitiTeknologiMalaysia,81310Skudai,JohorBahru,Johor,Malaysia

dWirelessCommunicationsandNetworkingLab,DepartmentofElectronicEngineering,MacquarieUniversity,Sydney,Australia

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received30July2013

Receivedinrevisedform17February2014 Accepted11March2014

Availableonline13April2014

Keywords:

Biterrorrate Cloudflash Cloud-to-groundflash Interference Microwaveradiation Wirelesssystem

a

b

s

t

r

a

c

t

Inthisstudy,cloud-to-ground(CG)flashandintra-cloud(IC)flasheventsthatinterferewiththe trans-missionofbitsinwirelesscommunicationsystemoperatingat2.4GHzwereanalyzed.Biterrorrate (BER)andconsecutivelostdatagram(CLD)measurementmethodswereusedtoevaluateBERandburst errorfrom3tropicalthunderstormsonNovember27,28,and29during2012northeasternmonsoonin Malaysia.Atotalof850waveformsfromtheelectricfieldchangerecordingsystemwererecordedand examined.Outofthese,94waveformsofveryfinestructurewereselectedwhichmatchedperfectlywith thetiminginformationoftherecordedBER.WefoundthatbothCGandICflashesinterferedsignificantly withthetransmissionofbitsinwirelesscommunicationsystem.Theseverityoftheinterferencedepends mainlyontwofactorsnamelythenumberofpulsesandtheamplitudeintensityoftheflash.The inter-ferencelevelbecomesworstwhenthenumberofpulsesinaflashincreasesandtheamplitudeintensity ofpulsesinaflashintensifies.Duringthunderstorms,wirelesscommunicationsystemhasexperienced mostlyintermittentinterferenceduetobursterror.Occasionally,inthepresenceofveryintenseNBP event,wirelesscommunicationsystemcouldexperiencetotalcommunicationlost.InCGflash,itcanbe concludedthatPBPisthemajorsourceofinterferencethatinterferedwiththebitstransmissionand causedthelargestbursterror.InICflash,wefoundthatthetypicalICpulsesinterferedthebits transmis-sioninthesamewayasPBPandmixedeventsinCGflashandproducedcomparableandinsomecases higheramountofbursterror.NBPhasbeenobservedtointerferethebitstransmissionmoreseverely thantypicalICandCGflashesandcausedthemostseverebursterrortowirelesscommunicationsystem.

©2014ElsevierB.V.Allrightsreserved.

1. Introduction

Wirelesscommunicationsystemhasevolvedrapidlyfromthe lastdecadeandutilizedwiderangeoffrequencybandsfortheir operationparticularlyinthemicrowaveregion.Currently, broad-bandwirelesssystems(e.g.WiFi,WiMAX)areoperatingbetween 2and6GHzfrequencybands.Recentobservationsofmicrowave radiationfromlightningflashesin[1,2]havetriggeredinterestto

Correspondingauthorat:ÅngströmLaboratory,DivisionforElectricity,

Depart-mentofEngineeringSciences,UppsalaUniversity,Box534,S-75121,Sweden. Tel.:+46184715849;fax:+46184715810.

E-mailaddresses:riduan.ahmad@angstrom.uu.se,riduan@utem.edu.my

(M.R.Ahmad),monariza.esa@angstrom.uu.se(M.R.M.Esa),

Vernon.cooray@angstrom.uu.se(V.Cooray),eryk.dutkiewicz@mq.edu.au

(E.Dutkiewicz).

studyitseffectsonwirelesscommunicationsystem.Theworkin [1]hasobservedstrongmicrowaveradiationat1.63GHz associ-atedwithcloud-to-ground(CG)flasheventssuchaspreliminary breakdownprocess(PBP),steppedleaders(SLs),dartleaders(DLs), andreturnstrokes(RSs).Moreover,similarmicrowaveradiation at2.4GHzassociatedwithcloud(IC)flasheventssuchasnarrow bipolarpulses(NBPs)wereobservedin[2].

Work donein[3] pioneeredthe investigationontheeffects of interferencefrom naturallightning flashes onwireless com-municationlinksat2.4GHz.Areceivedsignalstrengthindicator (RSSI)measurementmethodhadbeenusedwith1mlineofsight (LOS)separation.ThemeasuredRSSIvalueswereusedtosimulate biterrorrate(BER)fordata,audioandvideotransmissions.The measurementwasconductedinMalacca,Malaysiaduringarainy dayinSeptember 2010.Itwasconcludedthat lightningflashes reducethereceivedpowerandleadtounreliablecommunication links.

http://dx.doi.org/10.1016/j.epsr.2014.03.022

(2)

In[4,5]theeffectsofinterferencefromnaturallightningflashes onBERof anaudiotransmission wererecordedat 2.4GHz and 5.8GHz.Themeasurement wasconductedin Malacca,Malaysia betweenJanuaryandMarch2011.ThehighestrecordedBERwas 9.9×10−1withthemeanBERwas2.07×10−2.Thereference mea-surements under fair weather conditions give a baseline mean BERvalueat1.75×10−5.BERvaluesduringthunderstormsthat higherthan1.75×10−5wereinterpretedasbeinginterferedbythe lightningflashes.Thereforeitwasconcludedthatwireless commu-nicationsystemoperatingat2.4GHzwassignificantlyinterferedby lightningflashes.Howeverduetothelackoflightningflashevents information,thetypeofCGandICeventsthathaveinterferedwith thebitstransmissioncouldnotbeidentified.

Furtherworkin[6]evaluatedpossiblebursterroreventtooccur duringthebitstransmissionofanaudiostreamingapplicationat 2.4GHz.Aconsecutivelostdatagram(CLD)measurementmethod wasusedtoevaluatethebursterroreventduring3heavy thunder-stormsbetweenJanuaryandMarch2011inMalaysia.Itwasfound thatthenumberoflostpacketsperburstforallthunderstormswas rangingfrom8to40lostpackets.Howeverduetothelackof light-ningeventsinformation,whichCGandICeventscontributedtothe bursterroreventcouldnotbeidentified.

Inthispaper,wearemotivatedtoinvestigatewhattypeofCG andICeventsthatpossiblyhaveinterferedthebitstransmission ofawirelesscommunicationsystemoperatingat2.4GHz,which couldnotbeidentifiedby[3–6].Moreover,itisinterestingtostudy whichtypeofCGandICeventswouldinterferemoreseverelythan theothersandcontributetothehighBERandbursterrorevents. RecordingfromtheBERmeasurementsystemandtheelectricfield changerecordingsystemhavebeensynchronizedtoprovide com-montimestampinformation.Lightningwaveformstogetherwith theBER and maximum CLD from3 tropical thunderstormson November27,28,and29during2012northeasternmonsoonin Malaysiawereexaminedandanalyzed.

2. Measurementcampaign

Themeasurements wereconducted betweenNovember and December2012duringthenortheastmonsoon periodat Obser-vatory Station in Universiti Teknologi Malaysia (UTM), Johor, Malaysia,locatedatsouthernpartofpeninsularMalaysia(1◦ N,

103◦E).Themeasuringsystemwassituatedonatopofahillthat

is132mabovesealevelandabout30kmawayfromTebraustrait. AverticallypolarizedwhipantennawasconnectedthroughRF coaxialcabletoaLeveloneWUA-0614wirelessnetworkcard[7] attachedtoalaptopactingasaserver.Thesamelaptopactedas aclientand wasconnectedtoaCiscoWUSB600Nwireless net-workcard[8]throughaUSBcable.Atthesametime,thelaptop actedasa GPS-basedtime synchronizationserverand hasbeen synchronizedwithdigitalstorageoscilloscope(DSO)oftheelectric fieldchangerecordingsystem.Thelaptopwasbattery-powered andlocatedinsidetheobservatorystationbuilding.

Boththetransmitting(whipantenna)andreceivingantennas werepositionedonaplasticstructure(polyvinylchlorideorPVC) 1mabove ground and 3mfrom the observatory building. The ‘ground’referstotheEarth’ssurface.A5mdistanceseparatedthe transmittingantennafromthereceivingsystem.Fig.1illustrates theseconfigurations.Boththetransmitterandthereceiverwere operatingat2.4GHz.Adaptivemodulationswerechosen.Taking 5mofLOSrangeintoconsideration,mostprobably64-quadrature amplitudemodulation(QAM)wasfullyutilized.

Theapplicationlayerattheserverofthetransmittingsystem emulatedaRealAudioapplicationbroadcastingaudiocontentfrom amultimediaCD-ROM.Theaveragesendingdataratewas80Kbps. Thesizeoftheaudiodatawas1MB.Thisdatawastransmittedusing

Fig.1.BERandCLDrecordingsystemlocated3mfromtheobservatorystation building.Thetransmittingantenna(verticallypolarizedwhipantenna)was con-nectedtoaserverlaptopthroughthecoaxialcable.Thereceivingsystemwas connectedtoaclientlaptopthroughtheUSBcable.Boththetransmittingand receiv-ingantennaswerepositionedontopofaplasticstructure(PVC)1mfromground andseparatedby5mdistance.

realtimeprotocol(RTP)[9]overuserdatagramprotocol/Internet protocol(UDP/IP)asillustratedbyFig.2in[5].Thepayloadtype oftheRTPwasG.729[10,11].Thetotaloverheadaddedtoasingle audiodatapacketwas40bytes(fromRTP,UDPandIPheaders). HoweverdatagramsegmentationattheUDPlayerandpacket frag-mentationattheIPlayercouldincreasemorethetotaloverhead addedtotheaudiodatapacket.Thusthetotalnumberofbytessent fromtheserverwas1MBaudiodataincludingthetotaloverhead dependingonhowmanydatagramswasgenerated.

(3)

TheLevelonewirelessnetworkcardforwardedthesepackets tothereceivingsystemoverthewirelesscommunicationlinkby using802.11nradiointerface[12]withamaximumsendingrateof 150Mbps.Distributedcoordinatedfunction(DCF)protocol with-outtherequest-to-send/clear-to-send(RTS/CTS)mechanismwas chosenfortheoperationof802.11nradio.Thehandshaking mech-anismwasdisabledafterthehiddennodeproblemwaseliminated completely.Alsoothermeasurestoeliminatehiddennode prob-lemhadbeentaken suchasproviding strongLOSpath,shorter Transmitter–Receiverseparationandtheusageofomni-directional antenna.

The802.11nradiointerfaceatthereceivingsystemreceived thetransmittedbytes andforwarded themto theclientlaptop throughUSBinterfacewithamaximumdatarateof480Mbps. For-warderrorcorrection(FEC)mechanismwasusedinthe802.11n radiointerfacetocorrectanydetectederrorinthereceivedbytes. Theframechecksequence(FCS)attheendofeachframedetects most of the errors that are not corrected by the FEC scheme. TheseerrorsarecorrectedbyretransmissionsattheMAC layer. AnybitthatcannotbecorrectedbytheFECandFCSwascounted as an error at the IP layer.At the UDP layer, a packet with a certainnumberoferrorbyteswasconsidereddamagedand dis-carded. The report of the total number of bytes and packets in error was transmittedto the serverafter the run time was completed.

Theelectricfieldchangerecordingsystemconsistedofa paral-lelflatplateantennaplacedonametalstand1.5maboveground connectedviaashort60cmshieldedcoaxialcabletoa battery-poweredbuffercircuit. Theparallelflatplateantennastogether withthebuffercircuit(shieldedinametalcaseasshowninFig.2) were positioned about 9m away from the observatory station building.Thetopplateoftheantennawasconnectedtotheinner conductorofthecoaxialcablewhilethebottomplatewas con-nectedtothescreening conductorofthecoaxialcable.Alsothe bottomplatewasconnecteddirectlytothesinglepoint ground-ingsystem.Thebuffercircuitwasconnected toa LeCroyWave Runner44Xi-ADSOvia10mlongshieldedcoaxialcable.The con-figurationsofparallelflatplateantennaandbuffercircuitwere similartotheconfigurationsusedin[13].TheDSOdigitizedthe outputofthebuffercircuitat25MegaSamples/s(20nstime reso-lutionwith500msfullwindowsize)and8-bitverticalresolution witha100MHzbandwidth(10nsimpulsewidthresolution).The DSO wastriggered byeither positive edge or negative edge of incomingsignalwithtriggerlevelwasvariedbetween0.1Vand 1V.

3. Resultsandanalysis

Atotalof850waveformsfromtheelectricfieldchange recor-dingsystemwererecordedandexaminedduringthismeasurement campaign.Outofthese,94waveformsofveryfinestructurewere selectedwhichmatchedperfectlywiththetiminginformationof therecordedBER.78outof94arenegativeCG(-CG)flash wave-formsandtherestareICflashwaveforms.5outof16ICwaveforms areNBPevent.

EachBERpointcorrespondstothenumberoferrorbits mea-suredatthereceiverforacompletetransmissionofa1MBtrain ofbitsfromthetransmitter.Thetransmissiontimeforacomplete 1MBtrainofbitsis800ms.EachmaximumCLDpointcorresponds tothenumberofconsecutivelostdatagramobservedatUDPlayer. Itisanindicatorofhowseveretheoccurrenceofbursterrorin wirelesscommunicationsystem.Aslightningflashconsistsevents withimpulsiveactivities,itismorelikelythenatureofinterference willbeintheformofbursterror.

Thereferencemeasurementsunderfairweatherconditionsgive abaselineaverageBERvalueataround1×10−5.ThusBERvalues duringthunderstormsthathigherthan1×10−5areinterpretedas beinginterferedbylightningflashes.Ingeneral,acommunication linkforbroadcastaudiostreamsisrecommendedtomaintainBER below1×10−3[14].Inordertoincreasetheaccuracyofthe anal-ysisandfollowingtherecommendationoutlinedby[14],wehave chosenhigherbaselineBERvalueataround1×10−3.Therefore, lightningwaveformswhichhaveassociatedBERvalueshigherthan 1×10−3areregardedasthesourceofinterferencetothewireless

communicationsystem.Atfrequenciesbelow10GHz,attenuation byatmosphericgasesand rainmaynormallybeneglected[15]. Thus,weareconvincedthatthewirelesscommunicationsystem wasaffectedbythethunderstormalone.

3.1. Interferencefromcloud-to-groundflashes

ThreecategoriesofnegativeCG flashhave beenobserved to interfere the wireless communication systemsignificantly. The firstcategoryofCGflash(Category1)consistsoftypicalground flashwithoutPBPandcomeswithorwithoutchaoticpulses(CP). Fig.3 shows anexampleof 4 RSs(withassociated SLand dart steppedleader)withoutCP.Atotalof20waveformsarefallunder thiscategorywith8waveformsconsistCP.WeobservedthatallCP wereoccurredpriortosubsequentRSorinbetweenRS.Inaddition toRSandCP,thesecondcategoryofCG(Category2)consistsPBP

(4)

Fig.4.Activityrecordedon27November2012,tracenumber230at16:22:47,thetimewindowduringthismeasurementwas500mswithpre-triggerdelayof150ms.This typeofactivityisfallunderCategory2ofCGflash.

asillustratedinFig.4.Atotalof38waveformsarefallunderthis categorywith16waveformsconsistCP.

ThethirdcategoryofCGflash(Category3)consistsofmixed eventsoflargebipolarpulses(Fig.5a),trainofunipolarand bipo-larpulses(Fig.5b),andNBP(Fig.5c)thatweredetectedtogether withearliercategories.Atotalof20waveformsarefallunderthis category.AscanbeobservedfromFig.5,thelargebipolarpulses, NBPandtrainofpulseswerefoundtooccurbetweenand follow-ingRS.AlltheNBPeventswereobservedtohavelowerintensity comparedtothelargestRSbutcomparabletothelowestintensity RS.

Thepulseduration of thelargebipolarpulses and NBPwas observedtobelessthan100␮s.Thepulsedurationofthelarge bipolarpulsesisintherangebetween80and100␮swhileNBP

pulsedurationisintherangebetween5and20␮s.Incontrast,the durationofthetrainofunipolarandbipolarpulseswasobservedto exceed100␮s.Furthermore,wedetectedonlynegativepolarityof largebipolarpulsesandNBPcorrespondstotheBERvaluehigher than1×10−3.Ontheotherhand,thetrainsofpulsesweredetected withbothpolarities.

CorrelationofBERandCLDwiththenumberofRSisshownin Figs.6and7,and10forallnegativeCGflashcategoriestabulatedin Table1.KnowingthecorrelationofBERandCLDwiththenumber ofRS,onewillbeabletounderstandhowsignificantmicrowave radiationfromtheRSinterferingthetransmissionofbitsin wire-lesscommunicationsystem.Furthermore,oneisabletoidentify whichcategoryofCGinterferesmoreseverelythetransmissionof bitsandproduceshigherbursterror.Thebest-fitcorrelationlines

(5)

95% confidence limits Linear least sq. fit

Measured data Bit Error Rate

1.6

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0

−0.2

1 2 3 4 5 6

Number of return strokes

r2 = 0.85

y = mx + c m = 0.06

Fig.6.LinearcorrelationbetweenBERandthenumberofRSforCategory1ofCG flash.TheBERvaluehasbeengiveninpercentagescale,e.g.1×10−2correspondsto

1.

95% confidence limits Linear least sq. fit

Measured data

Number of return strokes

1 2 3 4 5 6

0 6

4

2

0

r2 = 0.93 y = mx + c m = 0.3

1 2 3 4 5 6

0 (A)

7

7 8 9

10

8

6

4

2

0 (B)

r2 = 0.89 y = mx + c m = 0.13 Bit Error Rate

Fig.7. LinearcorrelationbetweenBERandthenumberofRSfor(A)Category4a and(B)Category4b.TheBERvaluehasbeengiveninpercentagescale,e.g.1×10−2

correspondsto1.

Fig.8.Statisticaldistributionof(A)BERand(B)maximumCLDvaluesforCategories 1to3ofCGflash.TheBERvaluehasbeengiveninpercentagescale,e.g.1×10−2

correspondsto1.

Fig.9.Statisticaldistributionof(A)BERand(B)maximumCLDvaluesfor Cate-gory4ofCGflash.TheBERvaluehasbeengiveninpercentagescale,e.g.1×10−2

correspondsto1.

Table1

Definitionsandexplanationsofthecategoriesof-CGandICflashes.

Flash Category Events

-CG 12 RSsCategoryaccompanied1eventaccompaniedbysteppedleader,byPBPdartevent(stepped)priortheleaders,firstRSandchaoticpulses.

3a Category1eventaccompaniedbymixedeventsconsistslargebipolarpulses,trainofunipolarandbipolarpulses,andNBP. 3b Category2eventaccompaniedbymixedeventsconsistslargebipolarpulses,trainofunipolarandbipolarpulses,andNBP. 4a CombinationofCategory1andCategory3aevents,wherePBPeventisabsent.

4b CombinationofCategory2andCategory3bevents,wherePBPeventispresent.

IC 1 TypicalICpulses

(6)

95% confidence limits Linear least sq. fit

Measured data

Number of return strokes

1 2 3 4 5 6

Fig.10.LinearcorrelationbetweenmaximumCLDandthenumberofRSfor(A) Category4aand(B)Category4bofCGflash.

wereobtainedbyusingthemethodofleastsquaresregression.In thecasewhenlinearbest-fitlinecouldnotbeobtainedandthe cor-relationcoefficientrequirementcannotbesatisfied(r20.7),the

methodofnonlinearleastsquareswithGauss–Newton

optimiza-tionalgorithmwasappliedtoobtainthenonlinearbest-fit line. Thegradientvalue,mandinterceptvalue,cofthelinearbest-fit correlationlinescanbeobtaineddirectlyfromthefirstorderof polynomialequation,y(x)=mx+c.Ontheotherhand,thegradient value,bandinterceptvalue,aofthenonlinearbest-fitcorrelation linescanbeobtaineddirectlyfromtheequationofanexponential function,y(x)=aebx.

ForCategory1ofCGflash,theBERandthenumberofRSshows linearcorrelationwithr2=0.85andm=0.06asshowninFig.6.As thenumberofRSincreasesfrom1to6,theBERvalueincreases linearlyfrom1.5×10−3 to4.5×10−3.We can seeclearly from thefigurethat95%ofthemeasuredBERvaluesareliewithinthe range of confidence limitsof thefit line. Moreover,linear cor-relationisalsoobservedforCategory4aand Category4b ofCG flashinFig.7AandB,respectively.ThegradientofCategory4ais muchsmallerthanCategory4bwithm=0.13comparedtom=0.30, respectively,butmuchhigherthanthegradientofCategory1of CGflash,m=0.06.AsthenumberofRSincreasesfrom1to6,the BERvalueofCategory4bhasincreaseddrasticallyfrom1×10−2 to2.5×10−2,muchhigherthanCategory4aandCategory1with linearincrementfrom2×10−3to8×10−3andfrom1.5×10−3to

4.5×10−3,respectively.

Apparently, the BER performance of all CG flash categories becomesworstasthenumberofRSincreases.Further,this obser-vationmaysuggest that PBPin Category4b hasradiatedmore significantmicrowavepulsesthatinterferedwiththebits trans-mission,asareflectiontothemuchhighergradientandBERvalues thanCategory1andCategory4aofCGflash.Also,itcanbe sug-gestedthatthemixedeventsinCategory3ofCGflashradiated significantmicrowavepulsesinterferedwiththebitstransmission

95% confidence limits Nonlinear least sq. fit

Measured data (A) Bit Error Rate

(B) Maximum CLD

Intensity (Digizer volt) 10

Fig.11.Nonlinearcorrelationbetween(A)BERandthelargestintensity,and(B) maximumCLDandthelargestintensity,forCategory2ofCGflash.TheBERvalue hasbeengiveninpercentagescale,e.g.1×10−2correspondsto1.

95% confidence limits Nonlinear least sq. fit

Measured data (A) Bit Error Rate

(B) Maximum CLD

Intensity (Digizer volt) 10

(7)

Fig.13.Activityrecordedon27November2012,tracenumber257at16:51:26,thetimewindowduringthismeasurementwas500mswithpre-triggerdelayof150ms. ThistypeofactivityisfallunderCategory1ofIC.

butnotasmuchasPBP,asareflectiontolowerBERvaluesandsmall differenceingradientvaluesbetweenCategory1andCategory4a ofCGflash.

The statistical distribution of BER values in Fig. 8 supports thereflectionsofthecorrelationanalysis.Category1ofCGflash recordedthesmallestBER rangeandalsothelowestmeanBER valueat3.8×10−3whencomparedtoothercategoriesasevidenced inFig.8A.ThePBPinterferedthebitstransmissioninthemost sig-nificantwaycausesgreaterBERrangeandalsohighermeanBER valuesat1.77×10−2and3.58×10−2forCategory2andCategory 3bofCGflash,respectively.Category3arecordedonlyaslightly increaseinthemeanBERvaluewhencomparedtoCategory2of CGflash,2.25×10−2comparedto1.77×10−2,whichsuggeststhat theinterferencefromthemixedeventsisalmostcomparableto PBP.However,itisimportanttonoticethattherangeofBERvalues forCategory3aismuchsmallerthantherangeofbothCategory2

andCategory3bofCGflash.Moreover,therangeandmeanBERof Category4aaremuchsmallerthantherangeandmeanBERof Cat-egory4basevidencedinFig.9A.Therefore,itcanbesuggestedthat themixedeventshaveinterferedthebitstransmissionsignificantly butnotasmuchasPBPevent.Consequently,thismayexplainwhy themixedeventsinCategory3ofCGflasharesuggestedtoradiate significantmicrowavepulsesbutnotasmuchasPBPasrevealed fromthecorrelationanalysis.Further,wecouldobservethatallthe BERvalues(outliersexcluded)ofCategory1ofCGflashand Cate-gory4aarebelow1.5×10−2whilemorethan50%oftheBERvalues ofCategory2ofCGflashandCategory4barehigherthan1×10−2as evidencefromthemedianvaluesinFigs.8Aand9A.MostlikelyPBP eventisthestrongestsourceofmicrowaveradiationinCGflash.

AlinearcorrelationbetweenmaximumCLDandthenumberof RScanbeseenclearlyfromFig.10Awithr2=0.91andfromFig.10B withr2=0.94forCategory4aandCategory4b,respectively.Asthe

(8)

(A)

(B)

Intensity (Digizer volt) 10 Nonlinear least sq. fit

Measured data

Fig.15.NonlinearcorrelationbetweenBERandthelargestintensityfor(A)Overall CGflash,(B)TypicalICflash,and(C)NBP.TheBERvaluehasbeengiveninpercentage scale,e.g.1×10−2correspondsto1.

numberofRSincreasesfrom1to6,thenumberoflostdatagram forCategory4ahasincreasedfrom5to20datagrams.ForCategory 4b,thenumberoflostdatagramhasincreasedmoredrasticallyand higherthanCategory4a,from20to60datagrams.Moreover,the gradientofCategory4bisestimatedtobe7.8,muchhigherthanthe gradientofCategory4aat2.7.Therefore,itcouldbesuggestedthat Category4bradiatedthemostsignificantmicrowavepulses,which interferedwiththebitstransmissionandcausedthehighestburst error.Apparently,correlationanalysisrevealsthatthebursterror ismoreprevalentasthenumberofRSincreasesandPBPinterfered thebitstransmissioninthemostsignificantway.

Further,statisticaldistributionofCLDvaluesinFig.8Bsupports therevelationofcorrelationanalysiswherewecanobservethat Category3brecordedthehighestmeanCLDvalueat73.78when comparedtoCategory3aandCategory1with42.09and7.15mean CLDvalues,respectively.EvidencefromFig.9Brevealsthemean CLDofCategory4bisdoublethemeanCLDofCategory4a.Also, therangeandmedianvaluesoftheformercategoryaremuchlarger thanthelattercategory.Inaddition,wealsoobservethatalltheCLD values(outliersexcluded)ofCategory4aarebelow20datagrams whilemorethan50%oftheCLDvaluesofCategory4barehigher

95% confidence limits Nonlinear least sq. fit

Measured data (A) Maximum CLD

(B)

Intensity (Digizer volt) 200

Fig.16.NonlinearcorrelationbetweenmaximumCLDandthelargestintensityfor (A)OverallCGflash,(B)TypicalICflash,and(C)NBP.

than20datagramsandupto190datagrams.Evidently,PBPisthe strongestsourceofmicrowaveradiationinCGflashthatinterfered thebitstransmissionandconsequentlyincreasestheBERandCLD values.

CorrelationofBERandmaximumCLDwiththelargestintensity is shown in Figs. 11 and 12. The largest intensity here corre-spondstothelargestamplitudeofRSandreflectsthedistanceto thesourceofradiation.ThelargestRSintensityisassumedtobe inverselyproportionaltothedistancefromthesourceofradiation. Consequently,closeradiationsourcewouldrecordsmuchhigher intensityRSthanthedistantradiationsource.ForCategory2ofCG flashandCategory4b,theBERandthelargestintensityshowsvery goodexponentialcorrelationwithr2=0.91(Fig.11A)andr2=0.93 (Fig.12A),respectively.Also,themaximumCLDandthelargest intensityshowsverygoodexponentialcorrelationwithr2=0.91 (Fig.11B)fortheformercategoryandr2=0.91(Fig.12B)forthe lattercategory.ThegradientofCategory2ofCGflashisestimated tobe0.4forbothBERandmaximumCLD,thesameasthe gradi-entestimatedforCategory4b.Astheintensityincreasesfrom1 to5volts,theBERandmaximumCLDvaluesofbothcategories increasesfrom1×10−2to4×10−2andfrom20to80datagrams,

(9)

Fig.17.Statisticaldistributionof(A)BERand(B)maximumCLDvaluesforoverallCG,IC,andNBPflashes.TheBERvaluehasbeengiveninpercentagescale,e.g.1×10−2

correspondsto1.

severelywiththebitstransmission.Alsothebursterrorismore prevalentastheintensityincreases.

3.2. Interferencefromcloudflashes

TwocategoriesofIChavebeenobservedtointerferethe wire-lesscommunicationsystemsignificantly.ThefirstcategoryofIC consiststypicalcouldflashpulsesasshowninFig.13.Atotalof 11waveformsfallunderthiscategory.ThesecondcategoryofIC isaspecialtypeofICknownasNBP.Atotalof5waveformsfall underthiscategory.Outofthese,4areisolatedNBPand1isin thecategoryofmultipleNBP(showninFig.14).AlltheNBPinthis categorywereobservedwithlargeintensitycomparableandeven largerthanthelargestRS,unliketheNBPfoundtooccuraspartof negativeCGflash.Further,wedetected2positivepolarityisolated NBPcorrespondingtotheBERvaluehigherthan1×10−3andthe othershavetheoppositepolarity.

CorrelationofBERandmaximumCLDwiththelargestintensity isshowninFigs.15and16.Thelargestintensityherecorresponds totheamplitudesofthelargestICpulseandNBPandreflectsthe distancetothesourceofradiation.ThelargestICandNBP intensi-tiesareassumedtobeinverselyproportionaltothedistancefrom thesourceofradiation.Consequently,closeradiationsourcewould recordmuchhigherintensitythanthedistantradiationsource.The BERandthelargestintensityshowsverygoodexponential correla-tionwithr2=0.70(Fig.15B)andr2=0.99(Fig.15C)forthefirstand secondcategoriesofIC,respectively.Also,themaximumCLDand thelargestintensityshowsverygoodexponentialcorrelationwith

r2=0.71(Fig.16B)andr2=0.99(Fig.16C)forthefirstandsecond categoriesofIC,respectively.ForbothBERandmaximumCLD,the gradientofthesecondcategoryofICismuchhigherthanthefirst categoryofICandoverallCGflashcategory(Figs.15Aand16A),2.0 comparedto1.3and0.4,respectively.

Thisresultmaysuggestthat,astheimpulsiveactivityofICgets moreintense,itwouldinterferemoreseverelywiththebits trans-mission.Alsothebursterror ismoreprevalentas theintensity increases.Further,thegradientvaluestellusthattheNBPradiated themostintenseimpulsivemicrowavepulses,evenwhen com-paredtoallcategoriesofCGflash.Thestatisticaldistributionsof BERandCLDvaluesinFig.17AandBaresupportingthisfinding

wheretheNBPrecordedthehighestmeanBERandCLDvaluesat 4×10−2and91.60,respectively.Ontheotherhand,thefirst cat-egoryofICrecordedmeanBERandCLDvaluesat2.54×10−2and

53.36,respectively,whicharefoundtobehigherthanthe over-allCGcategorywithmeanBERandCLDvaluesat1.69×10−2and 31.92,respectively.

4. Discussionandconclusion

(10)

However,asdiscussedin[17],thedatapointsofspectral ampli-tudeabove10MHzfrequencyregionsweremostlyscattered.Thus, itwasrecommendedthatmoremeasurementsshouldbedonefor individualeventathigherfrequenciestoobtainreliableresultsto resolvetheambiguitiesinthespectralshape.

Inthelightofveryrecentfrequencyspectrumstudyin[1],PBP, SLandDLfromanegativeCGflashhavebeenobservedto gener-atetrainsofindividuallyresolvablemicrowavepulsesat1.63GHz band.TheamplitudeofthemicrowavepulsesproducedbyPBPwas observedtobemoreintensethanSLandDL.Ontheotherhand, thenegativeRSwasobservedtogeneratenoise-likemicrowave burstinsteadofindividualpulsesbutwithmoreintenseamplitude whencomparedtoPBP.However,themicrowaveburstofRSlasted onlyforafewhundredsofmicrosecondswhilemicrowavepulses ofPBPlastedlongerforseveralmillisecondsduration. Thismay explainwhytheinterferencelevelfromCGflashissignificantly higherduringthepresenceofPBPasevidencedinFigs.8and9.The typicalPBPtraindurationisbetweenafewmillisecondandtens ofmillisecondsandthus,mostlikelythateachpulsesinthePBP trainradiatesmicrowavepulsesandinterfereswiththebits trans-missionfordurationlongerthanotherCGflasheventssuchasRS. Therefore,inCGflash,itcanbeconcludedthatPBPisthemajor sourceofinterferencethatinterferedwiththebitstransmission andcausedthelargestbursterror.

InICflash,wefoundthatthetypicalICpulsesinterferedthe bitstransmissioninthesamewayasPBPandmixedeventsinCG flashandproducedcomparableandinsomecaseshigheramount ofbursterror.Ontheotherhand,NBPhasbeenobservedto inter-ferethebitstransmissionmoreseverelythantypicalICand CG flashesasevidencedinFigs.15–17.Recentworkin[2]revealed thatNBPhasproducednoise-likemicrowaveburstlastedfor sev-eralmicrosecondsat2.4GHzband.Itisinterestingtowonderhow asingleNBPpulsecouldproduceveryintensemicrowave radi-ationpulses that couldinterferethebitstransmission severely. Ifwe considerthe very recentproposal putforwardin [19] to betrue that NBPdischarge is the result of relativisticelectron avalanchesmechanismratherthanconventionalleaderdischarge. Then,thespectralamplitudeoftheresultantradiationfieldwould bepeakedataround1GHzasestimatedin[20].Thismayexplain ourfindingand provide logical reason whya single NBPpulse couldproduceveryintensemicrowaveburstat2.4GHzbandand caused themost severeburst errorto wireless communication system.

Acknowledgements

Thispaperisarevisedandextendedversionofcontribution[4] kindlypresentedbythefirstauthoratthe2012International Con-ferenceonLightningProtection(ICLP).Wewouldliketothankthe participantsfortheirhelpfulcommentsandsuggestions.Research workinthis paperwasfundedbyMinistryofHigherEducation MalaysiaandAnnaMaria LundinFoundation(Småland Nation). ParticipationofVernonCooraywasfundedbythefundfromthe B.JohnF.andSveaAnderssondonationatUppsalaUniversity.We wouldlike tothankAssoc.Prof. Dr.Zulkurnain Abd.Malek,Dr.

Azlinda,Ms.Zaini,Mr.KamyarandMr.Behnam,fromIVAT,UTM, Malaysiaforthesupportsduringthemeasurementcampaign.

References

[1]D. Petersen,W.Beasley,Microwaveradioemissionsofnegative cloud-to-groundlightningflashes,Atmos.Res.135–136(2014)314–321.

[2]M.R.Ahmad,M.R.M.Esa,V.Cooray,Narrowbipolarpulsesandassociated microwaveradiation,in:Proc.ofProgressInElectromagneticsResearch Sym-posium(PIERS),Stockholm,Sweden,August12–15,2013,pp.1087–1090.

[3]M.R.Ahmad, M.Rashid,M.H.A.Aziz,M.M.Esa,V.Cooray,M.Rahman,E. Dutkiewicz,Analysisoflightning-inducedtransientin2.4GHzwireless com-municationsystem,in:Proc.ofIEEEInternationalConferenceonSpaceScience and Communication(IconSpace),Penang,Malaysia,July12–13,2011, pp. 225–230.

[4]M.R.Ahmad,M.R.M.Esa,M.Rahman,V.Cooray,Measurementofbiterrorrate at2.4GHzduetolightninginterference,in:Proc.ofInternationalConference onLightningProtection(ICLP),Vienna,Austria,September2–7,2012,pp.1–4.

[5]M.R.Ahmad,M.R.M.Esa,M.Rahman,V.Cooray,E.Dutkiewicz,Lightning inter-ferenceinmultipleantennaswirelesscommunicationsystems,J.Light.Res.4 (2012)155–165.

[6]M.R.Ahmad,M.R.M.Esa,V.Cooray,E.Dutkiewicz,Performanceanalysisof audiostreamingoverlightning-interferedMIMOchannels,in:Proc.of Interna-tionalSymposiumonCommunicationsandInformationTechnologies(ISCIT), GoldCoast,Australia,October2–5,2012,pp.513–518.

[7]LeveloneWirelessNanoUSBNetworkAdapterWUA-0614Datasheet. Avail-able:http://global.level1.com/datasheet/WUA-0614-V2SPECEN.pdf

[8]Linksys Ultra Rangeplus Dual-band Wireless-N USB Network Adapter WUSB600N Specifications (Network Adapter), CNET Review Web Site. Available: http://reviews.cnet.com/adapters-nics/linksys-ultra-rangeplus-dual/4507-33807-32816737.html

[9]H.Schulzrinne,S.Casner,R.Frederick,V.Jacobson,RTP:ATransport Pro-tocolforReal-TimeApplications,IETFRFC3550Standard2003.Available:

http://tools.ietf.org/html/rfc3550

[10]Implementors’GuideforITU-TRec.G.729(2007):CodingofSpeechat8kbit/s UsingCS-ACELP,TelecommunicationStandardizationSectorofITUSeriesG: TransmissionSystemsandMedia,DigitalSystemsandNetworks2009. Avail-able:http://www.itu.int/rec/T-REC-G.Imp729-200911-I/en

[11]IEEEstandardforinformationtechnology:Telecommunicationsand infor-mationexchangebetweensystems:Localandmetropolitanareanetworks: Specificrequirements,Supplementtocarriersensemultipleaccesswith col-lisiondetection(CSMA/CD)accessmethodandphysicallayerspecifications: Physicallayerparametersandspecificationsfor1000Mb/soperationover 4-pairofcategory5balancedcoppercablingtype1000BASE-T,IEEEstandard 802.3ab1999.

[12]IEEE standard for information technology: Local and metropolitan area networks:Specificrequirements:Part11:WirelessLANmediumaccesscontrol (MAC)andphysicallayer(PHY)specificationsamendment5:Enhancementsfor higherthroughput,IEEEstandard802.11n2009.

[13]M.R.M.Esa,M.R.Ahmad,V.Cooray,Waveletanalysisofthefirstelectricfield pulseoflightningflashesinSweden,Atmos.Res.138(2014)253–267.

[14]Y.Chen,T.Farley,N.Ye,QoSrequirementsofnetworkapplicationsonthe Internet,J.Inform.Knowl.Syst.Manage.4(2004)55–76.

[15]RecommendationITU-RP.618-9:PropagationDataandPredictionMethods RequiredfortheDesignofEarth-SpaceTelecommunicationSystems, Radio-communicationSectorofITUSeriesP:RadiowavePropagation2009.Available:

http://www.itu.int/rec/R-REC-P.618-10-200910-I/en

[16]C.D.Weidman,E.P.Krider,M.A.Uman,Lightningamplitudespectrainthe inter-valfrom100kHzto20MHz,Geophys.Res.Lett.8(1981)931–934.

[17]D.M.LeVine,ReviewmeasurementsoftheRFspectraofradiationfrom light-ning,Meteorol.Atmos.Phys.37(1987)195–204.

[18]U.Sonnadara,V.Cooray,M.Fernando,Thelightningradiationfieldspectraof cloudflashesintheintervalfrom20kHzto20MHz,IEEETrans.Electromagn. Compat.48(1)(2006)234–239.

[19]V.Cooray,G.Cooray,T.Marshall,S.Arabshahi,J.Dwyer,H.Rassoul, Electro-magneticfieldsofarelativisticelectronavalanchewithspecialattentionto theoriginoflightningsignaturesknownasnarrowbipolarpulses,Atmos.Res. (2014)(inpress).

Gambar

Fig. 1 illustrates
Fig.type 3. Activity recorded on 29 November 2012, trace  number 84 at 14:45:50, the time window during this measurement was 500 ms with pre-trigger delay of 150 ms
Fig.type 4. Activity recorded on 27 November 2012, trace number 230 at 16:22:47, the time window during this measurement was 500 ms with pre-trigger delay of 150 ms
Fig.1102 to   correspondsofCGThevalue  been   in  scale,     e.g. percentage  given   (A)has BER and (B)maximum  CLDvalues for Categories18.Statisticaldistributionofto 3 flash.BER  × −1.
+5

Referensi

Dokumen terkait

dan untuk memudahkan melayani para konsumen tanpa mengganggu dalam pekerjaan, penggambaran aplikasi ini sama halnya dengan serangkaian proses-proses yang terurut dalam

Sehingga tidak jarang terjadi degradasi beion ka(ena pelaksanaan yang tidak sesuai dengan pengaturan ya,1g berlaku, rrloisal adanya retak, pelapukan beton dan korosi

Pihak Otoritas Pelabuhan; Untuk dapat masuk dan memulai pekerjaan perbaikan ka- pal, terdapat beberapa ketentuan yang harus di sepakati antara pe- milik galangan

Pemilikan tanah berdasar hukum adat meskipun tidak diakui sebagai hukum positif namun secara realita diakui keberadaannya oleh masyarakat dan seringkali dijadikan dasar (alas

Diharapkan dalam pembuatan alat ukur ini dapat membantu masyarakat dalam pemantauan kualitas air Danau Toba secara cepat, akurat dan real time dengan biaya

(4) menyusun strategi pengembangan pariwisata agar dapat meningkatkan pendapatan masyarakat dan pemerintah daerah di sekitar obyek wisata Danau Toba. Metode penelitian yang

[r]

KUMPULAN CERPEN SENYUM KARYAMIN KARYA AHMAD TOHARI: KAJIAN STILISTIKA DAN NILAI