ContentslistsavailableatScienceDirect
International
Journal
of
Electronics
and
Communications
(AEÜ)
j o ur na l ho m e p a g e :w w w . e l s e v i e r . d e / a e u e
Multilevel/AES-LDPCC-CPFSK
with
channel
equalization
over
WSSUS
multipath
environment
Hakan
Cam
a,
Osman
N.
Ucan
b,
Volkan
Ozduran
c,∗ aTurkishAirForceAcademy,34149Yesilyurt,Istanbul,TurkeybIstanbulAydinUniversity,FacultyofEngineeringandArchitecture,DepartmentofElectricalandElectronicsEngineering,Florya,Kucukcekmece,Istanbul,Turkey cIstanbulUniversity,FacultyofEngineering,DepartmentofElectricalandElectronicsEngineering,34320Avcilar,Istanbul,Turkey
a
r
t
i
c
l
e
i
n
f
o
Articlehistory:
Received13December2010 Accepted24March2011 Keywords:
AdvancedEncryptionStandard Lowdensityparitycheckcodes ContinuousPhaseFrequencyShiftKeying WSSUS
Channelequalization
a
b
s
t
r
a
c
t
Inthispaper,inordertoensuresecureandrobustcommunication,anewtypeofdataencryptionand errorcorrectionmechanismcalledMultilevel/AdvancedEncryptionStandard-LowDensityParityCheck Coded-ContinuousPhaseFrequencyShiftKeying(Multilevel/AES-LDPCC-CPFSK)iscarriedout.Here,we havechosenAESfordataencryption,LDPCcodesforerrorcorrectionandCPFSKformodulation.We haveevaluatederrorperformanceofthisschemeoverWide-SenseStationaryUncorrelatedScattering (WSSUS)multipathchannelswithchannelequalization.Wehavesimulated5-levelAES,2-levelLDPCfor 4CPFSKand16CPFSKoverWSSUSchannelsmodeledbyCooperationinthefieldofScience&Technology, Project#207(COST207).Wehaveconcludedthatourproposedstructurehaspromisingresultscompared toTurboCodesforallSNRvaluesinWSSUSchannels.
© 2011 Elsevier GmbH. All rights reserved.
1. Introduction
Encryptionanderrorcorrectionmechanismsarerequiredfor secureandrobustcommunication.AESalgorithmisasymmetric blockcipherprovidingsecuresensitiveinformation.128,192or 256bitsofdatablockscanbeprocessedbyAESusingkeylengths of128,196and256bits.Itisbasedonroundfunctionswhichare ByteSubstitution,RowShifting,ColumnMixingandKeyAddition [1–3].LDPCcodesarewellknownerrorcorrectioncodes,providing lowencoderanddecodercomplexity[4].Manyauthorshave inves-tigatedLDPCcodes;ZyablovandPinkster[5],Margulis[6],Tanner [7],MacKayandNeal[8],Wiberg[9].Thesestudieshavedefined thebasicprinciplesofLDPCcodes.Therealsocanbefoundmany studiesinliteraturewhichhaveexpandedtherangeofLDPCcode usagebyHaleyetal.[10],Cuietal.[11],Lubyetal.[12],Johnson [13],BingandJun[14],Abematsuetal.[15],NikandFekri[16].
Bandwidthefficiencyisoneofthemainperformancecriteria inmultipathchannels.CPFSK,aspecialkindofContinuousPhase Modulation(CPM)[17],providesagoodbandwidthefficiencyand biterrorperformance withits additionalmemoryunit. Various studieshavebeeninvestigatedaboutthedifferentaspectsofCPFSK bythefollowingauthors;Osman[18],Sakamotoetal.[19],Altay [20],ChengandLu[21].ContinuousPhaseEncoder(CPE)and Mem-oryless Mapper (MM) are two components of CPFSK. CPEis a
∗ Correspondingauthor.Tel.:+905358415755. E-mailaddress:volkan@istanbul.edu.tr(V.Ozduran).
convolutionalencoder.Itproducescodewordsequencesthatare mappedontowaveformsbyMM,whichcreatescontinuousphase signals.CPEprovidesnotonlypowerefficiencybutalsophase con-tinuity.
WSSUSisabasicmultipathchannelmodeltodescribethefading dispersivechannels.Ithastwomainparametersforcharacterizing offadingandmultipatheffects,whicharepropagationdelayand Dopplershift[8].WSSUSchannelscanbemodeledasin Coopera-tioninthefieldofScience&Technology,Project#207(COST207) withstandardprofilessuchasTypicalUrban(TU),BadUrban(BU) andHillyTerrain(HT).SinceCOST207isaseverefadingchannel, variouschannelequalizerssuchasLeastMeanSquares(LMS)and RecursiveLeastSquares(RLS)areemployed.
Inthispaper,anewjointscheme,“Multilevel/Advanced Encryp-tionStandard-LowDensityParityCheckCoded-ContinuousPhase FrequencyShiftKeying(ML/AES-LDPCC-CPFSK)”issimulatedover WSSUS multipath environment with channel equalization. The basic idea of using multilevel encryption and encoding is to partition information bits into several levels and encrypt and encode each level separately by AEScipher and LDPCencoder. Inthisapproach,“ML/AEScipher”encryptsinformationbits,then “ML/LDPCencoder”[22]encodestheseparallelbits,afterthat,the codedbitsareturnedintoserialtoparallelaccordingtothetype of modulation. CPEencodes thelast level of thesebits to pro-videphasecontinuity,then,MMmapsthecodedbitsintoM-ary CPFSKsignalsandthesesignalsaresenttochannel.CPFSK demod-ulator demodulatesthenoisysignalsatthereceiver side.Then, “Signalconstellationandprobabilitycalculationblock”processes 1434-8411/$–seefrontmatter © 2011 Elsevier GmbH. All rights reserved.
thesesignals,and oneandzeroprobabilitiesofreceivedsignals areevaluated.Ineverylevel“LDPCdecoder” decodesthese sig-nalsandinputbitsareestimatedfromthesebits.Finally,“ML/AES decryptionblock”decryptsthesebitstreams.
Organizationofthispaperis asfollows:In Section2,abrief overviewofAESalgorithmisgiven.ThenLDPCcodesandCPFSK modulationaredescribedinSections3and4,respectively. ML/AES-LDPCC-CPFSKsystemsareinvestigatedinSection5.Finally,error performanceoftheproposedschemeisdiscussedinSection6. 2. AdvancedEncryptionStandard
AES is known worldwide as Rijndael symmetric-key block cipherwhichdesignedbyJoanDaemenandVincentRijmen.The blocksizeofAESis128bits,andthekeylengthscanbe128,192or 256bits[3].Thekeysizedeterminesthenumberofroundstobe performed.Forinstance,forthekeysizeof128,192and256bits, thenumberofroundsare10,12and14,respectively.
Inourproposedscheme,weusetheversionofRijndaelthathas a128-bitkey,operateson128-bitplaintextsandhas10rounds. Multipleof128bits(5×128bits)arefedintoAESEncryptionblock. TheinputlevelofML/LDPCblockdeterminestheinputlevelofAES. Itisbothpossibletooperateonlongsizedatablocksandprovide bandwidth-efficienttransmissionatthesametimewithmultilevel inputs.
3. LowDensityParityCheckCodes
LDPCcodesarethemembersofbinarylinearblockcodesfamily. Thesecodescanbedescribedasverysparseparitycheckmatrix H,whichcontainsmanyzerosandonlyfewones.Itispossibleto representaLDPCcodeasagraph,namelybipartitegraph,whose node-setcanbepartitionedintotwonon-emptysetsVandCand everyedgeconnectsanodeinVwithanodeinC.Thevariableand checknodescorrespondtothecolumnsandrowsofparitycheck matrixHandrepresentabitsymbolinthecodewordsandaparity equationofcode,respectively.Anedgebetweeneachvariablenode andchecknodecanberepresentedbya“1”incorrespondingrow andcolumnintheparity-checkmatrix.Inthispaperonlyregular LDPCcodesareinvestigated.InregularLDPCcodes,everycolumnof Hcontainsjamountof“1”andeveryrowofHisfilledwithkamount of“1”,whichjandkdenotevariablenodedegreeandchecknode degree,respectively.TherearetwokindsofLDPCcodes:regularand irregular[23].HereregularLDPCcodesareinvestigated.Inregular LDPCcodesallrowsandcolumnsoftheparitycheckmatrixHhas thesamedegreewhichcorrespondstothenumberof“1”bitsin theserowsandcolumns.Allvariablenodeshavedegreejandall checknodeshavedegreekinabipartitegraphofregularLDPCcode. Thebipartitegraphofaregular(3,4)LDPCcodeisdepictedinFig.1 andthecorrespondingparitycheckmatrixofthiscodeisdescribed asanexample. H=
⎡
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎣
0 0 1 0 0 1 1 1 0 0 0 0 1 1 0 0 1 0 0 0 0 0 0 1 0 0 0 1 0 0 0 0 1 1 1 0 0 1 0 0 0 1 1 0 0 1 0 0 1 0 1 0 0 0 0 1 0 0 1 0 0 0 0 1 1 0 0 0 1 0 0 1 1 0 0 1 1 0 1 0 0 0 0 0 0 0 0 0 0 1 0 1 0 0 1 1 0 1 1 0 0 0 0 0 1 1 0 0⎤
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎦
(1) 3.1. EncodingofLDPCcodesLDPCcodes canberepresented withtwo parameters:(n,k), wherencorrespondstoinformationbitsandkcorrespondsto
code-Variable Node
Check Node
Fig.1.Bipartitegraphrepresentationofregular(3,4)LDPCcode.
words.ALDPCcodewithaparitycheckmatrixH,whichhasn−k rowsandkcolumns,encodeskinformationbitsintoncodewords. CodewordsofthisLDPCcodesatisfythefollowingrelation:
HxT=0 (2)
whereH,xandTreferstoparitycheckmatrix,codewordand trans-poseoperation,respectively.TheparitycheckmatrixHofanLDPC codecanbedefinedas:
H=[Hp|Hu] (3)
whereHpisparitybitssubmatrixandHuisinformationbits
sub-matrix.Likewisecodewordscanbedefinedas:
X=[Xp|Xu] (4)
wherexpisasetofparitybitsandxuisasetofinformationbits.
Thefollowingrelationcanbedefined:
b=HuXuT (5)
Withthisequality,encodingofabinarysystematic(n,k)LDPCcode becomesequivalenttosolvingthefollowingequation:
b=HpXpT (6)
Fromthisequationthefollowingrelationcanbedefined:
XpT=H−1p b (7)
3.2. DecodingofLDPCcodes
LDPC codes can be decoded by Message Passing Algorithm (MPA),which iterativelyupdates the probabilitiesof bitnodes.
Someauthorshavecalleddifferentnamesforthisalgorithmlike Sum Product Algorithm (SPA) or Belief Propagation Algorithm (BPA)atthesametime.Atthispoint,sometermsandformulations whichweusedinMPAaredefinedasfollows:
yi:setofreceivedsignal
ci:setofbitnode
b:0,1
qij(b):theprobabilityofmessagetobepassedfrombitnodecito
thechecknodefj
rji(b):theprobabilityofmessagetobepassedfromchecknodefj
tothebitnodeci
Rj=setofcolumnlocationsofthe1sinthejthrow
Rj/i: setofcolumnlocationsof1sinthejthrow,excludingthe
variablenodei
Cj=setofrowlocationsofthe1sintheithcolumn
Ci/j:setofrowlocationsof1sintheithcolumn,excludingthecheck
nodej
Kij:constantvalues
Qi(b):posteriorprobabilityofthecodebitci
pi=prob(ci=1|yi)
DecodingprocessofLDPCCodescanbeexplainedstepbystep inthefollowingsection.
Step1:Initialization
Allbitnodessendtheirqij(0)andqij(1)messages.Sincenoother
informationisavailableatthisstep,theprobabilitiesofmessages (qij(0)andqij(1))canbecomputedusingEqs.(8)and(9).
qij(0)=1−pi=prob(ci=0|yi)= 1 1+e−2yi/2 (8) qij(1)=pi=prob(ci=1|yi)= 1 1+e2yi/2 (9) Step2:Paritynodeupdates(firsthalfrounditeration)
Thechecknodescalculatetheirrji(0)andrji(1)response
mes-sagesaccordingtoEqs.(10)and(11). rji(0)= 1 2+ 1 2
i∈Rj/i (1−2qij(1)) (10) rji(1)=1−rji(0) (11)Step3:Bitnodeupdates(secondhalfrounditeration)
The bitnodes update theirresponse messages to thecheck nodes.So,theprobabilitiesofmessages(qij(0)andqij(1))canbe
updatedusingEqs.(12)and(13). qij(0)=Kij(1−pi)
j∈Ci/j rji(0) (12) qij(1)=Kijpi j∈Ci/j rji(1)) (13)whereconstantskijareselectedtoensureqij(0)+qij(1)=1
Step4:Softdecision
AllthebitnodescalculatetheQi(0)andQi(1)posterior
proba-bilitiesofthecodebitciusingEqs.(14)and(15).
Qi(0)=Ki(1−pi)
j∈Ci rij(0) (14) Qi(1)=Kipi j∈Ci rij(1) (15)whereconstantskiareselectedtoensureQi(0)+Qi(1)=1
Step5:Harddecision
Finally,allthebitnodesupdatetheircurrentestimationci of
theirvariableci bycalculatingtheprobabilitiesfor0 and1
andvotingforthebiggeroneasinEq.(16). ˆci=
1 Qi(1)> 1 2 0 other (16)
Step6:If ˆcHT=0orthenumberofiterationsreachesthemaximum
limit,thenthealgorithmterminates,otherwiseitreturnstostep2. 4. ContinuousPhaseFrequencyShiftKeying
CPFSKisaspecialkindofCPM[23],whichhasanM-dimensional form.Atthispointsometermsandformulationswhichareusedin CPFSKaredefinedasfollows:
(t,Y):tilted-phaserepresentationofCPM
Y:inputsequenceofM-arysymbols,Yi∈{0,1,2,...,M−1}.
h: modulationindex h=J/P,where Jand Parerelativelyprime integers
T:channelsymbolperiod q(t):phaseresponsefunction L:integer
g(t):frequencypulse s(t):transmittedsignal
f1:asymmetriccarrierfrequencywheref1=fc−h(M−1)/2T
fc:carrierfrequency
Es:energyperchannelsymbol
0:initialcarrierphase
Tilted-phaserepresentationofCPMwasderivedbyRimoldiin [17],withtheinformation-bearingphasegivenby
(t,Y )=4h
∞
i=0
Yiq(t−iT ) (17)
Here,Jisgenerallychosenas1andPiscalculatedas2tothepower ofthenumberofmemoriesinCPE.Modulationindexcanbe com-putedfromtheequationofh=1/P.Thephaseresponsefunction q(t)isacontinuousandmonotonicallyincreasingfunctionsubject tothefollowinginequalities:
q(t)=
0 t≤0
1/2 t≥LT (18)
Thephaseresponseiscanbedescribedasin
q(t)=
t
−∞
g()d (19)
Transmittedsignals(t)canbecomputedfromthefollowing equa-tion:
s(t,Y )=
2Es
T cos(2f1t+(t,Y )+0) (20) wheref1Tisassumedasanintegerforsimplificationwhenusing
theequivalentrepresentationoftheCPMwaveform. 5. Wide-SenseStationaryUncorrelatedScattering enviromentandequalization
Indigitalcommunicationsystemsmultipathfadingisthemost importantfactorforsystemperformance.Duetovariousobjects surrounding,scattering,diffractionandreflectioncausethe multi-pathfadingwhichhasfrequencyandtimedispersivenature.Since
Fig.2.GeneralMultilevel/AES-LDPCC-CPFSKscheme.
changingcharacteristicsofchannelduetothemovementofthe mobilestationitcanbeassumedasatime-variantandmultipath fadingchannel and this channel canbe described as statistical WSSUS channel. Thischannel model canbe characterizedby a twodimensionalscatteringfunctionwhichhastwoparameters; Dopplerfrequencyduetothemobilemovementandechodelay duetomultipatheffects.Fadingstatisticsofthischannelmodelare assumedtobeconstantforashorttimeinterval.Sincesevere trans-missionconditionsofWSSUSchannelsandtheaddednoise,signals arecorruptedandneedtoberegeneratedbyusingchannel equaliz-ers.Undersevereconditionsreceiversrequirethecharacteristicsof atransmissionchannelforchannelidentification.Butthese charac-teristicsarenotalwaysavailableandchannel’sequivalentimpulse responsehastobeestimated.Itisimpossibletoestimate infor-mationbitswithoutchannelequalization.Thuschannelequalizer blockisnecessaryforallschemesinWSSUSchannelsfor perfor-manceimprovement.
6. Proposedscheme
InFig.2,blockdiagramofdesignedjointML/AES-LDPCC-CPFSK modelisdepicted.Ascanbeseenfromthis figure,thereexists a “ML/AESEncryption block”for theencryption of messagebit sequences.Multilevelinputstothisblockaremultipleof128bits (M×128bits)andaredependentontheinputlevelof“ML/LDPC block”[22].Aftertheencryptionblock,thereisaLDPCencoderat everylevelofthesystemmodelandaCPEwhichisserially con-nectedtotheLDPCencoderatthelastlevel.Encodedmessagebit sequenceisconvertedfromserialtolog2Mparallelbranchinthe
multilevelschemeaccordingtothetypeofM-aryCPFSK modula-tion.Then,eachLDPCencoderprocessestheinformationsequence simultaneously.Thelevelof“ML/LDPCblock”determinesthelevel of“ML/AESEncryptionblock”(i.e.2,3and4levelsofLDPC,the lev-elsofAESare5,7and10,respectively).TheoutputofthelastLDPC encoderisrunthroughtheCPE.CPEisusedtoshapethemodulated signal’sspectrumforphasecontinuity.Finally,allencoderoutputs andCPEoutputaremappedtoCPFSKsignalsbyamemoryless map-peraccordingtopartitioningrule.Inthepartitioningrule,signal setisdividedintotwosubsetsbytheMSBofSerialtoParallel(S/P) converteroutput.Ifthefirstoutputz11is0,thenthefirstsubset
ischosen,ifz11is1,thenthesecondsubsetischosen.Thesecond
outputz12bitdividesthesubsetsintotwogroupsasthesimilar
way.Thispartitioningprocesscontinuesuntilthelastpartitioning level.Thesesignalsarerunthroughthechannel.Noiseisadded accordingtothechannelmodel.ChannelEqualizerblockprocesses thesecorruptedandnoiseaddedsignalsequence.Atthereceiver sideofthecommunicationchannel,“CPFSKdemodulator” demod-ulatesthecorruptednoisysignals.Then“Signalconstellationand probabilitycalculationblock”processesthesesignalstoevaluate oneand zero probabilitiesof receivedsignals.After that“LDPC decoder”decodesthesesignalsandinputbitsareestimatedfrom decodedbitsineverypartitioninglevel.Finally,thesebitstreams aregonethroughthe“ML/AESdecryptionprocess”andestimated bitsequencescanbedecrypted.Forabetterbiterrorperformance, someofthemessagebitsequenceswhichwillbeknownbyreceiver blockasinpilotsymbolcommunication[24],canbeusedinthelast levelofAEStoestimatechannelparameters.
In Fig. 3, the partitioning mechanism for 4CPFSK system is depictedinorder toexplainthepartitioning of ML/AES-LDPCC-CPFSK. In thefirst partitioninglevel z1 defineswhich subset is
chosen.Ifz1=0,thenfirstsubset{s0,s1,s4,s5}ischosen,ifz1=1,
thenthesecondsubset{s2,s3,s6,s7}ischosen.Atthesecond
par-titioninglevel,ifweassumethatthefirstsubsetischoseninfirst partitioninglevel,z2z3bitsspecifywhichsignalwillbetransmitted
tothechannel.Similarly,inthefirstpartitioninglevel,ifthesecond subsetischosen,z2z3bitsspecifywhichsignalwillbetransmitted
tothechannel.Table1summarizesthisprocess.
ItcanbeseenfromFig.3thatinitialandendingphaseof trans-mittedsignalwilltake(0,)valuesifmodulationindexhofan LDPCC-4CPFSKsystemischosenas1/2.Input–outputdataand sig-nalconstellationsforthissystemaresummarizedinTable1.Here,
Table1
Input–outputandsignalconstellationfor4CPFSK.
n ˇn z1z2z3 n+1 4CPFSK 0 0 000 0 s0 0 1 001 s1 0 2 010 0 s2 0 3 011 s3 0 100 s4=−s0 1 101 0 s5=−s1 2 110 s6=−s2 3 111 0 s7=−s3
01
0
0
{s
0,s
2}
{s
1,s
3}
{s
4,s
6}
{s
5,s
7}
θ
nθ
n+1π
π
z1=0 z1=1s
0s
4s
5s
1s
7s
3s
6s
2z
2z
3=
00 01 10 11 00 10 11 s0 s1 s5 s4 s2 s3 s7 s6Fig.3.4CPFSKh=1/2signalphasediagram.
z1issystematicbit,z2andz3arethebitsencodedbyCPE.Here,z2
helpstoshowusatwhichphasethesignalwillstartatthenext
codinginterval.Iftheinitialphase(n)is“0”and“z2”is0,the
end-ingphaseoftheinstantsignalandthestartingangleofthenext
signalphase(n+1)is“0”,ifn=0andz2=1,n+1=,ifn=1and
z2=0,n+1=,ifn=1andz2=1,n+1is0.Iftheinitialphaseis“0”,
thenthesignalispositive,iftheinitialphaseis“1”,thenthesignal
isnegative.Onlyiftheseconditionsaremetiscontinuitygranted.
AccordingtoFig.3andTable1,thetransmittedsignalsatthephase
transitionsareasfollows:from0phaseto0phase,s0,s2;from0
phasetophase,s1,s3;fromphaseto0phase,s5,s7andfrom
phasetophase,s4,s6.
Messagepassingalgorithmisusedfordecodingatthereceiver sideofLDPCcodes.Oneandzeroprobabilitiesofreceivedsignal areusedinthisalgorithm.Foreverydecodinginterval,these prob-abilitiesarefirstlyevaluatedforallparallelinputbranchsequences
accordingtopartitioningrule.Atthispoint,sometermsand formu-lationsaredefinedasfollows:
rk:receivedsignal
L:partitioninglevel
si:transmittedM-aryCPFSKsignal
Pji:probabilitywhereidenotes0or1
Oneandzeroprobabilitiesofreceivedsignalarecomputedfrom thefollowingequations:
PL 0= 1 M
M−1 i=0 1 (rk−s2i)2 (21) PL 1= 1 M M−1 i=0 1 (rk−s2i+1)2 (22)ThenreceivedsignalismappedtoonedimensionalBPSKsignalas canbedescribedinbelow.
L=1− 2·P L 0 PL 0+PL1 (23) InFig.3,itcanbeshownthatineverypartitioninglevel proba-bilitycomputationsandmappingareexecutedaccordingtosignal set.TheseprobabilitiesarecalculatedasinEqs.(24)–(27)for par-titioninglevel1and2.
P01=14
3 i=0 1 (rk−s2i)2 (24) P1 1= 1 4 3 i=0 1 (rk−s2i+1)2 (25) P2 0= 1 4 1 i=0 1 (rk−si)2 + 1 (rk−si+4)2 (26) P12=14 1 i=0 1 (rk−si+2)2 + 1 (rk−si+6)2 (27)Fig.5.FivelevelAEStwolevelLDPCC-16CPFSKscheme.
Theseoneandzeroprobabilitiesofreceivedsignalareevaluated andthen ˆciisestimatedfromEqs.(9)–(13)foreverypartitioning
level.
Partitioningmechanismfor16CPFSKsystemhasthesame prop-ertyasin4CPFSK.
6.1. Five-levelAEStwo-levelLDPCcoded4CPFSK
In Fig. 4, five-level AES two-level LDPC coded 4CPFSK sys-temmodelisdepicted.Firstofall“5-levelAESEncryptionblock” encrypts5parallelmessagebitsequenceswhicharemultipleof 5×128bitsandthus5×128bitslonginputdatacanbeprocessed simultaneously.Five-levelinputs aredependentonthenumber oftwo-levelLDPC.Aftertheencryptionprocess,two-levelLDPC encoderencodestheseencryptedparallelbitsequences.Atthis stage, LSB bits are encoded by CPEfor phase continuity. Then memorylessmapper mapsthesecodedbitsinto4CPFSKsignals. ThesesignalsaresenttotheWSSUSmultipathchannel.Noiseis addedaccordingtothechannel model.ChannelEqualizerblock processesthesecorruptedandnoiseaddedsignalsequence.Atthe receiversideofthecommunicationchannel,“CPFSKdemodulator” demodulatesthenoisycorruptedsignals.“Signalconstellationand probabilitycalculationblock”is thenprocessedthesesignals in everyleveltoevaluateoneandzeroprobabilitiesofreceived sig-nals.Afterthat“LDPCdecoder”decodesthesesignalsandsoinput bitsequencesareestimatedfromdecodedbitsinevery partition-inglevel.Inthefinalstep,“5-levelAESdecryptionprocess”decrypts theseestimatedbitsequences.
6.2. Five-levelAEStwo-levelLDPCcoded16CPFSK
InFig.5,five-levelAEStwo-levelLDPCcoded16CPFSKsystem modelisdepicted.Inthisscheme“AESEncryptionblock”and“LDPC Encoderblock”worklikein“Five-levelAEStwo-levelLDPCcoded 4CPFSK”asdescribedinSection6.1.Thenencodedbitsareturned intoserialtoparallelwithtwobranchesforeachoutputof“LDPC Encoderblock”.Here,LSBbitsareencodedbyCPEagainforphase continuity.Thenmemorylessmappermapsthesecodedbitsinto 16CPFSKsignals.ThesesignalsaresenttotheWSSUSmultipath channel.Noiseisaddedaccordingtothechannelmodel.Channel Equalizerblock processesthesecorruptedand noiseadded sig-nalsequence.Atthereceiversideofthecommunicationchannel,
“CPFSKdemodulatorblock”,“Signalconstellationandprobability calculationblock”,“LDPCdecoderblock”and“5-levelAES decryp-tion block” work like in “Five-level AES two-level LDPC coded 4CPFSK”.Finally,estimatedbitsequencescanbedecrypted. 7. Performanceanalysisanddiscussion
Here, we investigate multilevel AES encryption, multilevel LDPCcoded CPFSK structures and join theseschemes as “Mul-tilevel/AdvancedEncryptionStandard-LowDensityParityCheck Coded-ContinuousPhaseFrequencyShiftKeying (ML/AES-LDPCC-CPFSK)”. Thus a real communication environment is achieved employingbothdataencryptionandchannelencoding.Westudy ontwodifferentmodulationtechniques;“FivelevelAEStwolevel LDPCC-4CPFSK”and“FivelevelAEStwolevelLDPCC-16CPFSK”over WSSUSchannelswithRLS,LMSequalization.Weuseregular(3,4) LDPCcodeswithablocklengthof302bits,andmaximum100 iter-ations.TheBitErrorRatio(BER)versusSignaltoNoiseRatio(SNR) curvesoftheproposedsystemsareobtainedforBU,TUandHT typesofCOST207(Figs.6–8).Inoursimulationweuseseventeen
-5 -4 -3 -2 -1 0 1 2 10-4 10-3 10-2 10-1 100 Eb/No (dB) BER
Five-level AES two-level LDPC coded 4CPFSK&16CPFSK over BU type of COST207
4CPFSK RLS 4CPFSK LMS 16CPFSK RLS 16CPFSK LMS
Fig.6. Performanceof five-levelAES two-levelLDPCcoded 4CPFSK&16CPFSK schemeoverBUtypeofCOST207channel.
-5 -4 -3 -2 -1 0 1 2 10-4 10-3 10-2 10-1 100 Eb/No (dB) BER
Five-level AES two-level LDPC coded 4CPFSK&16CPFSK over TU type of COST207
4CPFSK RLS 4CPFSK LMS 16CPFSK RLS 16CPFSK LMS
Fig.7. Performanceof five-levelAES two-levelLDPCcoded 4CPFSK&16CPFSK schemeoverTUtypeofCOST207channel.
channelcoefficientsforBU,TUandHTascanbeseenfromTable2. Weassumethevelocityofthemobileradiocommunication ter-minalas100km/h.ThecarrierfrequencyofCOST207channelis takenas950MHz.ThemaximumDopplerShiftistakenas88Hzand minimumDopplerperiodischosenas11.4ms.Finally,thesymbol periodisassumedas3.7s.
Althoughidealdataencryptionisassumedinmostoftherelated comparedmodels[18,20,23,24]weincludedataencryptionblock. WehavesimulatedourschemesforvariousSNRvaluesoverWSSUS channelmodels.AsitcanbeseenfromTables3and4,FivelevelAES twolevelLDPCC-4CPFSK&16CPFSKsystemsshowhigherBER per-formanceinallSNRvaluesover2Level-TurboCodes4PhaseShift Keying(2L-TC4PSK)[24]forTU,BU,HTtypeofCOST207withRLS, LMS.Thus,bothhighercodinggainandreductioninthenumberof CPFSKlevelsareachieved.
-4 -2 0 2 4 6 8 10-4 10-3 10-2 10-1 100 Eb/No (dB) BER
Five-level AES two-level LDPC coded 4CPFSK&16CPFSK over HT type of COST207
4CPFSK RLS 4CPFSK LMS 16CPFSK RLS 16CPFSK LMS
Fig.8. Performance offive-levelAES two-levelLDPC coded 4CPFSK&16CPFSK schemeoverHTtypeofCOST207channel.
Table2
ChannelcoefficientsofBU,TUandHTtypesofCOST207.
BU TU HT
−0.0484−0.2943i −0.1651+0.5534i −1.0607+0.2118i −0.0000−0.4282i −0.0506+0.7480i −1.3775−0.0015i 0.0447−0.5339i 0.0849+0.9113i −1.5213−0.1868i 0.0815−0.6063i 0.2169+1.0292i −1.4957−0.3254i 0.1064−0.6414i 0.3215+1.0907i −1.3252−0.4056i 0.1162−0.6368i 0.3778+1.0881i −1.0507−0.4232i 0.1090− 0.5919i 0.3702+1.0178i −0.7242− 0.3814i 0.0839−0.5077i 0.2910+0.8806i 0.4014−0.2902i 0.0418−0.3872i 0.1403+0.6815i −0.1342−0.1641i −0.0152−0.2351i −0.0733+0.4294i 0.0356−0.0203i −0.0835−0.0577i −0.3338+0.1373i 0.0820+0.1237i −0.1582+0.1375i −0.6196−0.1790i −0.0021+0.2523i −0.2335+0.3418i −0.9057−0.5011i −0.2047+0.3541i −0.3030+0.5460i −1.1663−0.8090i −0.4962+0.4224i −0.3599+0.7407i −1.3771−1.0820i −0.8341+0.4562i −0.3978+0.9165i −1.5177−1.3001i −1.1686+0.4600i −0.4108+1.0649i −1.5736−1.4452i −1.4497+0.4420i
Table3
ComparisonofBER-SNR(indB)valuesof2L-TC4PSK[24]andfivelevelAEStwolevelLDPCC-4CPFSKaforTU,BU,HTtypeofCOST207withRLS,LMS.
TU BU HT
RLS LMS RLS LMS RLS LMS
BER [24] a Gain [24] a Gain [24] a Gain [24] a Gain [24] a Gain [24] a Gain
1.00E−01 2 −2 4 0 −2 2 3 −2 5 4 −2 6 5 −2 7 7 −2 9 1.00E−02 3 0 3 4 0 4 5 0 5 6 0 6 7 1.8 5.2 9 1.8 7.2 1.00E−03 4 1.2 2.8 5 1.2 3.8 6 1.2 4.8 7 1.2 5.8 9 4.4 4.6 10 3.4 6.6 1.00E−04 5 2 3 6 1.6 4.4 7 1.6 5.4 8 1.6 6.4 10 8 2 11 4 7 1.00E−05 7 8 1.9 6.1 8 9 1.8 7.2 12 12 4.4 7.6 [24]:2L-TC4PSK.
aFivelevelAEStwolevelLDPCC-4CPFSK.
Table4
ComparisonofBER-SNR(indB)valuesof2L-TC4PSK[24]andfivelevelAEStwolevelLDPCC-16CPFSKaforTU,BU,HTtypeofCOST207withRLS,LMS.
TU BU HT
RLS LMS RLS LMS RLS LMS
BER [24] a Gain [24] a Gain [24] a Gain [24] a Gain [24] a Gain [24] a Gain
1.00E−01 2 −5 7 0 −5 5 3 −5 8 4 −5 9 5 −5 10 7 −5 12 1.00E−02 3 −3 6 4 −2 6 5 −3 8 6 −3 9 7 −3 10 9 −2 11 1.00E−03 4 −2 6 5 −1.3 6.3 6 −2 8 7 −1.5 8.5 9 −2 11 10 −0.5 10.5 1.00E−04 5 −1.85 6.85 6 −1.2 7.2 7 −1.85 8.85 8 −1.3 9.3 10 −1.85 11.85 11 −0.2 11.2 1.00E−05 7 −1.8 8.8 8 −1 9 8 −1.8 9.8 9 −1.2 10.2 12 −1.8 13.8 12 0 12 [24]:2L-TC4PSK.
Asitcanbeseenfromsimulationresults,ML/AES-LDPCC-CPFSK schemeshavebettererrorperformancethan2L-TC4PSK[24] mod-els.
8. Conclusion
Inthispaper,wepresentanewjointencryptionanderror cor-rectionstructuretoensuresecureandrobustcommunicationin WSSUSchannelswithoutadditionalcomplexity.Forthispurpose, weintroduceMultilevel/AdvancedEncryptionStandard-Low Den-sityParityCheckCoded-ContinuousPhaseFrequencyShiftKeying andweevaluateerrorperformanceoverWSSUSchannelsforTU,BU andHTtypeofCOST207withRLS,LMSequalization.Itisshownthat ourproposedsystemprovidessecure,reliabledatatransmission withhigherrorcapability,bandwidthefficiencyandreductionof transmissionpowerusage.Thus,oursystemhasimportantpower andbandwidthadvantagesandisverysuitableforlowpowerand bandlimitedapplications.
References
[1]DaemenJ,RijmenV.TheblockcipherRijndael.SmartCardResearchand Appli-cations,LNCS1820.Berlin:Springer;2000.pp.288–296.
[2] Daemen J, Rijmen V. AES proposal: Rijndael AES algorithm submission, September3,1999.
[3]FIPS197.AdvancedEncryptionStandard,FederalInformationProcessing Stan-dard(FIPS).Washington,D.C.:NationalBureauofStandards,U.S.Department ofCommerce;November26,2001[Publication197].
[4]GallagerRG. Low-densityparity-checkcodes. Cambridge,MA: MITPress; 1963.
[5]ZyablovV,PinksterM.Estimationoftheerror-correctioncomplexityofGallager low-densitycodes.ProblPeredInform1975;11(January):23–6.
[6]MargulisGA.Explicitconstructionofgraphswithoutshortcyclesandlow den-sitycodes.Combinatorica1982;2(1):71–8.
[7]TannerR.Arecursiveapproachtolowcomplexitycodes.IEEETransInform Theory1981;IT-27(September):533–47.
[8]MacKayDJC,NealRM.NearShannonlimitperformanceoflowdensityparity checkcodes.ElectronLett1996;32(August):1645–6.
[9]Wiberg N.Codes and decoding ongeneral graphs. Dissertationon. 440. Linköping,Sweden:DeptElect.LinköpingUniv.;1996.
[10]Haley D, Gaudet V, Winstead C, Grant A, Schlegel C. A dual-function mixed-signalcircuit forLDPCencoding/decoding.IntegrationVLSIJ2008, doi:10.1016/j.vlsi.2008.09.006.
[11] CuiY,SiX,ShenY.AnovelalgorithmofconstructingLDPCcodeswithgraph theory.In:CIS.2008.
[12] LubyMG,MitzenmacherM,ShrokrollahiMA,SpielmanD,StemannV.Practical loss-resilientcodes.In:Proc29thAnnuACMSympTheoryofComputing.1997. p.150–9.
[13]Johnson SJ. Burst erasure correcting LDPC codes. IEEE Trans Commun 2009;57(March(3)).
[14]BingD,JunZ.Designandoptimizationofjointnetwork-channelLDPCcodefor wirelesscooperativecommunications.ICCS2008.
[15]AbematsuD,OhtsukiT,KashimaT,JarotSPW.LDPCcodesforhighdatarate multibandOFDMsystemsover1Gbps.PACRIM’07.
[16]Nik HP, Fekri F. Results on punctured low-density parity-check codes and improved iterative decoding techniques. IEEE Trans Inform Theory 2007;53(February(2)).
[17]RimoldiBE.AdecompositionapproachtoCPM.IEEETransInformTheory 1988;34(March):260–70.
[18]OsmanO.Blindequalizationofmultilevelturbocoded-continuousphase fre-quencyshiftkeyingoverMIMOchannels:researcharticles.IntJCommunSyst 2007;20(January(1)):103–19.
[19]Sakamoto T, Kawanishi T, Izutsu M. Continuous-phase frequency-shift keying with external modulation. IEEE J Sel Top Quantum Electron 2006;12(July/August(4)).
[20] AltayG.Performanceofsystematicdistance-4binarylinearblockcodeswith continuousphasefrequencyshiftkeyingoverMIMOsystems.WirelessPers Commun2008;44:403–13,doi:10.1007/s11277-007-9364-2.
[21]ChengCC,LuCC.Space-timecodedesignforCPFSKmodulationover frequency-nonselectivefadingchannels.IEEETransCommun2005;53(September(9)). [22]LimpaphayomP,WinickKA.Powerandbandwidth-efficientcommunications
usingLDPCcodes.IEEETransCommun2004;52(March(3)):350–4.
[23] HekimY,OdabasiogluN,UcanON.Performanceoflowdensityparitycheck codedcontinuousphasefrequencyshiftkeying(LDPCC-CPFSK)overfading channels.IntJCommunSyst2007;20:397–410.
[24] UcanON,BuyukatakK,GoseE,OsmanO,OdabasiogluN.Performanceof multilevel-turbocodeswithblind/non-blindequalizationoverWSSUS mul-tipathchannels.IntJCommunSyst2006;19:281–97.