Anda di halaman 1dari 10

2002 Int. Thammasat J. Sc.Tech.,Vol.7, No.

3, September-December

PAPR Reductionin OFDM Systems


and W.A.C. Fernando P. Foomooljareon Asian Instituteof Technology, Program, Telecommunications Thailand. 12120, Pathumthani P.O.Box 4, Klong Luang, -5130 4 1, T el.+66-2-524-51 Fax.+66-2-524 EmaiI : femando(4)ait.ac.th

Abstract modulation DivisionMultiplexingor OFDM is a form of multi-carrier Frequency Orthogonal to High spectralefficiency,robustness channelfading, immunityto impulseinterference, technique. and lessnonlineardistordensity,capabilityofhandling very strongechoes spectral uniform average of of tion are amongthe favoriteproperties OFDM. Eventhoughthereare manyadvantages OFDM, it offset.In this pahigh Peakto AveragePowerRatio (PAPR)and frequency hastwo main drawbacks: to the are Two new algorithms proposed reduce PAPR. per,the issueof PAPR in OFDM is discussed. properlyusing a lookuptable and the by selecting input sequences The first algorithm is carriedout to for beforethey are transformed the time dothe secondby scalingthe input envelope subcarriers main by InverseFast Fourier Transform(IFFT). Simulationresultsshow that the PAPR can be reducedsignificantlywith boththe schemes. PowerRatio Division Multiplexing(OFDM), Peak-to-Average Frequency Keywords: Orthogonal (PAPR), QPSK.

Introduction and of thanthirtyyears research Aftermore


carried out in different places, developments orthogonal frequency division multiplexing in (OFDM) has beenwidely implemented high Due digital communications. to the recent speed (DSP) processing in advancements digitalsignal circuits (VLSI) very large scale integrated and of the technologies, initial obstacles OFDM implementations not exist any more. Meando while, the use of Fast Fourier transform(FFT) arraysof sinusoidal has eliminated algorithms required generators and coherentdemodulation the andmade implemenparallel datasystems in cost effective.In recent tation of the technology in yearsOFDM has gaineda lot of interest diThis applications' versedigital communication properties like high hasbeendueto its favorable to spectralefficiencyand robustness channel fading.Today,OFDM is rnainlyusedin digital system(DAB) initiatedby audio broadcasting CCETT in France[1], and digitalvideo broadan (DVB) [2] enabling end-tosystem casting which is specsystem, end digitaltransmission trally efficient and ruggedagainstchanneldis-

tortions.This can be used for servicessuch as capacityfor program HDTV, offering increased serial data In broadcasting. the conventional transmissionsystem,the information symbols where eachsymbol sequentially are transmitted occupiesthe entire available spectrum bandthe width. But in an OFDM system, information and to is converted N parallelsubchannels sent division multiat lower ratesusing frequency plexing. The subcarrierfrequency spacing is selectedcarefully such that each subcarrieris zero crossing locatedon the other subcarriers' points.This impliesthat there is overlapping with but amongthe subcarriers will not interfere at eachother,ifthey are sampled the sub carrier are This meansthat all subcarriers frequencies. orthogonal. such as The OFDM has many advantages to high bandwidth efficiency, robustness the fading problem,use of small guard selective interval, and its ability to combatthe ISI probis equalization needed lem. So, simplechannel channelequalizainstead complexadaptive of tion.

70

2002 Thammasat J. Sc.Tech.,Vol.7, No.3, September-December Int.

of Apart from variousadvantages OFDM, there also. The frequency are certain disadvantages and offset ofthe subcarriers the high PAPR are of the major drawbacks OFDM [3]. ln this paper,the PAPR problemhas been are and addressed two new algorithms proposed to reducethe PAPR. The rest of the paper is provides The next section organized follorvs. as followedby a to an introduction OFDM systern, 'uvork on of detailed analysis PAPR,the related and PAPR reduction,the proposedschemes and results. Finally,the conclusions simulation the scopefor futurework concludes paper. Introduction to OFDM of An OFDM symbolconsists N subcarriers by the frequencyspacingof Al Thus, the B totalbandwidth will be dividedinto N equally are And all the subcarriers subcarriers. spaced orthogonalto each other within a time interval can be of lengthT =11 Lf. Each subcarrier with the complex modulated independently modulationsymbolX,,n, where m is a time inindex.Then within the dex and n is a subcarrier time interval Z the following signalof the z-th by OFDM block periodcan be described equat i o n( 1 ) a s :
t I N-l sr,,

LZr^.,sn| \ t ' m = on = o N o w , c o n s i d ea s i n g l e F D M s y m b o l( m : 0 ) r O without loss of generality.This can be shown because there is no overlaDbetweendifferent OFDM symbols.

= x{r) f

o N - l

-mT)(3)

can Sincelz = 0, X,n.n be replacedby X" .Then, as the OFDM signal canbe described follows:.

= .r(r) -!

1/-l

i ^ t ! jv

f x,niz^trt
H

(4\

r=0

If the bandwidth of the OFDM signal is by B = N x L/ andthe signal x(t) is sampled the 1 | s a m p l i n t i m eo f A r = = e . t h e nt h eO F D M B NLf signal is in discretetime form and can be writ(5). ten as shownin equation . N-l
H

,, =l\
' -

t , rl l^ y l

x , u i ' n o ' ' Nk = 0 , 1 , . . . , N - l ,


(5)

n=0

xr(l)=

* LI ^.r8r(t ! ' " n= o

-mt ) \t)

(2). where, g,(r) is definedthroughequation I exp(i2nnLft\. I u. 0<t<T e$e

g,(r)=1^

Q)

where, g,(t) is a rectangularpulse applied to each subcarrier[4]. The total continuoustime of signal x(r) consisting all the OFDM blocksis (3). givenby equation

where, n denotesthe index in frequencydomain andX, is the complex symbol in frequency domain. Furthermore, equation (5) can be expressed usingthe IFFT [5]. Figure I showsa typical systemblock diagram of an OFDM system. The serialinput data to bit streamis converted Nparallel subchannels and mappedwith a selectedmodulation scheme (signalmapper), conresultingin Nsubchannels taining information in complex number form.

2002 Thammasat J. Sc. Tech.,Vol.7, No.3, September-December Int.

t.
OFDM system Figure 1: BasicFFT based Thesecomplex valuesare then sent to the 1y'channelIFFT. Then, the parallel signalsare by convertedback to a serial sequence using a P/S device. A guard interval is insertedto reduce the effect of ISI caused by multipath propagation. Finally, the signal is convertedto an analog signal and convertedback up to a At a form suitablefor transmission. the receiver, reverse procedure is used to demodulatethe only a simOFDM signal.In an OFDM system, a ple equalizer, more specifically, singletap or equalizer,is neededat the receiverto remove the ISI. Peak-to-Average Power Ratio (PAPR) As explained earlier, one of the major of drawbacks OFDM is the very high peak-toaveragepower ratio (PAPR). PAPR of OFDM with the number subof exponentially increases carriers.If power amplifiers are not operated to with large power back-offs,it is impossible power below the specified keepthe out-of-band limits. This situationleadsto very inefficient so transmitters, it is amplificationand expensive the to highlydesirable reduce PAPR. analysis mathematical a In this section, detailed The Root Mean Square for PAPR is presented. (RIVIS)magnitudeof the OFDM signal is deofthe envefined as the root ofthe time average ( V P 1.where P is defined equaby lopepower tion (6). where, x(r) is the OFDM signal defined by ( e q u a t i o n 3 ) . T h e v a l u e F i n t h i s c a s ec o r r e sponds a singleOFDM symbol,and depends to of on the sequence informationcarryingcoeffipower of OFDM symcients{X, } . The average bols can be written asP*=E{F}. fnus, the canbe defined as, PAPRofan OFDM signal

max.lx(r)12
"
t

tel},Tl
p

(7) maxlxu)12 r.[o,r)

n\ xTtl'l
the input data power is normalized, t | , .. . t 2 \l f t l . a n dw e g e t . t h e nt l l x ( / ) l i= If
,

r l

max lx(r)l re[o,r)

...t2

(8)

(e) ,=,ffi1# Yr,"''^onl'


I

lrr r
t r ! - r

<-llt
< .^y'

Nl-^

y _' t l z m n J t l
I
I

, l

12

(10)

ln=U

(11)

ar= P= : ltxtnl2 lZlx,l' ' 1 / T J '


t=0 n=0

, T

, N - l

,u,

It clearly shows that the maximum PAPR is N. equal to the number of subcarriers For MPSK modulation. .thereare only ,Dl sequences

72

2002 Thammasat J. Sc. Tech.,Vol.7, No.3, September-December Int.

in havingmaximumPAPR N as described [6]. that the This means, numberof sequences gives very high PAPR is not very high. If the number increases, the ratio of the seof subchannels quence that givesvery high PAPR and all disrapidly. The overall decreases tinct sequences for numberof distinct sequences the ly' subcarwith M-PSK is,4.y'. Thusthe rier-OFDMsystem (12) and (13) by ratio can be obtained equation

,"- M, MN
R = 142-N

(r2)
(13)

Related Works Severalalgorithms[7-13] have been proHowever, posedto handlethis PAPR problem. signifialgorithms haveproduced noneof these of cantreduction PAPR in OFDM systems. (PTS) was proPartial Transmit Sequence bearing posedin [7]. In PTS, the information block is subdividedinto disjointed subcarrier rotationfactors and introduced carriersubblocks for each sub-blockand modified the subcarrier amplitudevector. Thereby,PAPR was reduced with different rotationfactorsfor different subto of This needs number iterations find a blocks. the optimum combination of factors for subblocks. Adaptive PTS [8] was proposedto reduce the number of iterationsby setting up a desiredthresholdand trial for different weighing factors until the PAPR droppedbelow the we the With this approach can reduce threshold. numberof iterationsand the complexityof the of by system only 0.1% lossin reduction PAPR. with are considered In [8], 256 subcarriers Resultsshowedthat the PAPR can be QPSK. by reduced 4.1 dB and 4.0 dB withoutadaptive HowPTS and with adaptivePTS respectively. need to send side ever these two approaches informationto the receiverwhich implies a reductionin the bandwidthefficiency. side information[9] is PTS with embedded another approachthat can be combined with both conventionaland adaptivePTS. This approach embedsa combinedknowledgewithin the transmitteddata. so no extra bits are sent.

word errorsduringdetection But theseintroduce of the sequence information. A simple Encodable/Decodable OFDM in code QPSK proposed [10] usedReed-Muller with QPSK.This couldreduce PAPRto less the than 6 dB but it could not be usedwith higher ordersignalconstellations. OFDM PAPRreduction by a rotationof redundancy position in bit subblock code word schemewas proposedin the bit [11]. In this method redundant positions of subblockcodewords are rotatedand the lowest PAPR codeword is chosenby a feed back for scheme. However, the side information bit position required. is as Oversized IFFT [2] is proposed another scheme to solve this problem. In oversized IFFT, clippingand filtering are doneby forward and inverseFFT. This can avoid out of band power but with somein band distortion,overall and shrinkingof constellation with the introducCompandtion of somenoiselike components. ing transform[i3] compresses large signal a while enhancing smallsignalthat can achieve a a desiredPAPR but with a significantincrease in the bit error rate(BER). Proposed Methods to Reduce PAPR I. PAPR Reduction using Lookup Tables We have studiedthe effect of BPSK and schemes PAPR, for syson QPSK modulation tems with different number of channels (2^ numberof channelsin the set), and found that to QPSK is a betteroption as compared BPSK. the Table I summarizes informationaboutthe 0 found in the study with dB PAPR sequences QPSK as the modulationformat. The study is and furlimitedto a maximumof 16 channels the idea to ther work is in progress generalize for and methodology any numberofchannelsof the form 21. Using the informationgiven in Table 1, we can find an expressionfor the number of sequences eachcaseas follows. in then Let ft be an integer, : Numberof channels 2l inputsequenc "r: Numberof all possible Number of = zx+
L

42r

sequenceswith 0 dB PAPR

2002 Int. Thammasat J. Sc. Tech.,Vol.7, No.3, September-December

Table 1: The numberof QPSK input sequences that gives0 dB PAPR


A

Number of channels

Numberof all distinctinput sequences l6 256 65.536 4,294,967,296

Number.Of sequences which 0 dB PAPR output 8

2
4

2
l 4

32

r28
512

t6

Now, consider 8 channelsystemof Table l. the the We can generate 0 dB PAPR OFDM system and informationrate of l116, with 8 subcarriers by groupingthe input bit streaminto a groupof 6 or 8 bits. The outputof the QPSK signalmapper is groupedin either3 or 4 complexnumbers to (eachcomplexnumbercorresponds a symbol) with table containand eachgroup is compared to corresponding the ing 0 dB PAPR sequences is sequence sentto the 8 The matching system. c h a n n eIlF F T .

sr.iali

penX.l

>

tnl

Lo4kilL\

Trbb

i
selection Figure 2(a): Block diagramfor the input sequence

S a!i

QPSi( seq. ? SYrbo-irc'Prc (5 syo bob C oil Phx nunber' .. Y


: n ddFrt

-'-"! )i
,
b d 4 q , !

Gr o u po f 5 QP s K s y i b o l
V {5 s'rtbolrl"."r!, "> 5dUiisa"".{

S equen ce se bctrlg

t,ookup i
tabb I

\.

8On*"""",
V I ditbb".r.> 8 QPSx synbob :

,_--__.--.__lr

5 dbtofbase4 .

num ber' V d.c


I

i c roup of8 QPsK synbob


trrl i

5rib jish!.r => {dbtaddcss

:
D e c i na l n n g e 0 ' 1 0 23
erd j

Figure 2(b): Flowchartof the lookuptableapproach Figure 2 shows the block diagramof the proposedalgorithm.In figure 2(a), channel I to i are related to the number of real information and the rest containresidualvaluesto channels make the sequence0 dB PAPR. As proved on above,i can be either 3 or 4 depending the Thus, the inserial complex number sequence. formationrate is log(128)/8, which is equal to 7/16. Since most of the practical systemscan afford to have a PAPR of more than 0 dB, we

74

2002 Int. Thammasat J. Sc.Tech.,Vol.7, No.3, September-December

the can increase informationrateuntil the PAPR the reaches maximumtolerablelevel of the sysin of tem.Thenthe number inputsequences the as lookuptablecan be increased shownin Table flowchart the 2. Figure2(b) presents complete PAPR' to tableapproach reduce of the lookup
Table 2: Maximum PAPR and information rate lnformation I Maximum Numberof rate useful PAPR (average Input value) (dB) sequences

0
0.91 |.'76

t28
| 152 21'71

3.5 5.0
5.5

1t16
518

The complementarycumulative distribuhas tion function(CCDF = Pr(PAPR> PAPR6)) PAPR of an OFDM sysbeen usedto measure our tem. We also useCCDF to present results. Figure 3 shows CCDF distributionfor the purFor lookup table approach. the comparison of the original OFDM syspose,CCDF curves tem and the PTS with M = 2 havealso beenincluded.It clearly showsthat the PAPR can be reducedto 0.97 dB with the proposedscheme; the whereas PTS can reducethe PAPR only up to 4 dB. However,the informationrate of the is lookup table approach lower than that of the PTSscheme. II. PAPR Reduction using EnvelopeScaling We have studied the effect of envelope the scalingby scalingfactorsto reduce PAPR in the OFDM system.In our study, we use 256 technique and subcarriers the QPSK modulation inputs of the envelopes all the subcarrier so that ideais to scalethe enare equal.The proposed to of velope the inputin somesubcarriers obtain the minimumPAPR at the outputof the IFFT' The final input that givesthe lowestPAPR will be sent to the system.Note that this input sequencehas the samephaseinformationas the are original one,but the envelopes different.So the receiver can demodulatethe received sewithoutanysideinformation. quence The detailed explanationis given as fol4, lows.In Figure the inputN datablock is partiFirst or tioned into disjointsubblocks clusters. definethe datablock as a vectoras: x =lxo Xl Xr-r]' Then partition

tv16 28140 5' 7180 62180


67180

2.32
2.68

2433 2945 5504


12096

5.6 5.'7 6.2


6.'7

2.92 I 3.0

of An addedadvantage this methodis that it can detect possibleerrors as the system and transmitsonly a specific set of sequences out ofthis setcanclearlybe any sequence hence guess be can as concluded an error.An educated errorcorrection' to considered be a coarse Simulation Results

--*
+.' Lookup Original M-2 -r-PTS

e , , ,

LE-2

by X into M disjoint sets,represented the vec= 1 . 2 . . . .M l . A l l v e c t o rX ^ a r e s . r o r s t x , nm then scaled with the scaling fact<lrs r . . ( O , t ] , m = 7 , 2 , . . . M a n d c o m b i n e dt o getherbeforebeingsentto the IFFT.
M

x' = \srX,
m=1
0 2 4 8 6 PAPRo(dB) 10 12

(14)

x' = IFFT\X' j

(15)

of Figure3: Comparison CCDFof PAPRwith lookuptableandPTSwith M: 2

The scaling factors are chosen to minimize the PAPR ofx'. A sub-optimum iteration is in-

75

2002 Int. Thammasat J. Sc.Tech.,Vol.7, No.3, September-December

troducedto reducethe complexityof the calculations. The scalingfactorsare set to the one digit d e c i m as u c ha s 0 . 4 ,0 . 5 . . .0 . 8 a n df o r e x a m p l e l let all the scalingfactorsare {0.5, 1}. After set into M clusters, dividing the input sequence the all scalingfactors,s-= 1, calculate PAPR s o f t h e c o m b i n e d i g n a la s s h o w n i n e q . ( 1 4 ) . f t N e x t .c h a n g eh e f i r s l s c a l i n g a c t o r( s ' = 0 . J ) PAPR. If the new the and recompute resulting PAPR is lower than that in the previousstep. otherwise retain s, as paft of the final sequence.

changeit back to the one that gives the lowest PAPR (in this caseit is 1). The algorithmcontinues in this fashionuntil all sm are explored caseas explained in general in the particular but above,only M iterationshave been done.Note values s^canbe morethan of that the possible in 0.5 and I as explained this examplebut it increasesthe system complexity. Figure 5 Scaling showsthe flow chart of the E,nvelope alsorithm.

X, X2 Serialto Parallel and Data I X > Pattition Source , I into r Clusters

cluster lenoth l

ls'

Ctusterllnoth

F,.....

'I's,

.'

+
>.,(,r > lS'
r i"" '' l t

>

N pornts IFFT

l
ia

ootimization

scaling Figure 4: Block Diagramfor envelope

2002 Int. Thammasat J. Sc.Tech.,Vol.7, No.3, September-December

r-****--\ sert

\-.--.---___,..--.r
I

- v sq. td Cdalde PAPR Org.


? inp.{ seq X io M clGg x^ ;m = 1..M lruffimby1

"' >im<=M

'

.. YeS l:'

>

Mtltidyx,wihs i=1..M

P*lilim

No

V IFFT

s" =" (ot)


V M'Jtidy X, wilh S i r 1..M V

Y PAPRd = Ors. seq. PAPR S =t;m=1..M

+
PAPRffi T PAPR

_i\

v .. ..Etd ' /i

CfwEe bacf S*tt 1 t No

Yes

IFFT

PAPR<= PAt'Rm' <

V PAPR cdcridio

scaling Figure 5: Flow Chartfor envelope efficiently.


-0
j

SimulationResults

a4naj

i I

+rr$ --* r

I =8 i lacbtr0l I hcsF0j l , . i , ^ i i l

I l J-i i

- *.- facbFlJ

l j
I

I i I l I
I

i t

8 8

PAPRo(dB)

PAPRo(dB)

of Figure 6: Comparison CCDF of PAPRwith scalingfactor= 0.5 and PTS with M= 8 cumulative In Figure 6, the complementary distribution function (CCDF : Pr(PAPR > PAPR6)) of an OFDM signal with envelope is scaling approach shownwith 1/: 256 subcarriers, and QPSK as the modulationschemefor The resultsin Figure 6 are all the subcarriers. for varying the numberof clusterM with fixed scalingfactor at 0.5. It showsthat by increasing the number of M PAPR can be reducedmore

of Figure 7: Comparison the resultwith PTS and one scalingfactorvalue Figure 7 represents CCDF distribution the with differentscalingfactors.The value of M is fixed at 256 clusters equalto Nand variesthe or scaling factor. As comparedto the modified PTS [14] with M -- 8, the loh PAPR is reduced to by 1.8dB and by 4 dB as compared the original symbol.Note that the smallerscalingfactor sivesmorereductionin PAPR.

2002 Thammasat J. Sc. Tech.,Vol.7, No.3, September-Deceriber Int.

--e-Prs
+ESA

i l

t.1,-01

-.+--. PTSwith TC ! **-- ESAwith'IC l

1.U-02
L

rd l.E-03

?
L

l.fi-04

r.Ii-05

Ur'No(dB)
i { u ,**,*,

of scaling Figure 9: BER Performance envelope (N=256,scaling factor:0.5) with PTS. Besidesthe significantreductionof the PAPR, the main advantage this approachis that the of receiver does not need any side information. However the complexity of the proposed for is scheme higherthanthe PTSespecially the approach can caseof M:N. But a sub-optimum be found if we need to reducethe complexity. is scheme that it The iimitation of the proposed modulation cannotbe appliedwith non-constant schemes. Conclusions two schemes to In this paper,we proposed reducethe PAPR in an OFDM system.In the is first approachthe input sequence compared with the sequence givesthe lowestPAPR in that the lookup table. Resultsshow that PAPR can be reduced 0 dB when the informationrate is to 1116 and to 0.97 dB when rate is 5/8, for an OFDM system with 8 channelsusing QPSK we modulation.In the secondscheme scalethe input sequence envelopeusing different scaling factors before transforming the new sequence into the time domainby IFFT. Resultsshowthat the PAPR can be reducedsignificantly,at most 4 dB for 1% CCDF without the needof any side informationfor the receiver.Finally the system with singlescalingfactor (0.5) and the number is equalto the numberof subcarriers of clusters recommended. Future work is required to extend the first algorithm to a systemwith large as number of subchannels well as to analyze other modulation techniquesand develop an

Figure 8: CCDF of PAPR with multi scaling factors Figure 8 shows the result for the caseof multipfe scalingfactorswhen M is fixed at 256 factorof0.5. It is clearthatthe perat a scaling of formance the multiple scalingfactorsis lower than that for the case of fixed scaling factor. power of the systemis considNow the average ered. The adaptivePTS approachis to change so only the phaseof the sequence, the average power of the OFDM symbol remainsthe same. we For envelopescaling approach, changethe makingthe outof envelope the input sequence put of the correspondingOFDM symbol to change. of Figure 9 shows the BER performance the OFDM system with envelopescaling and the PTS with and without Turbo Coding for a As fadingchannel. we comfrequency selective of pare only the performances the two PAPR scalingand PTS, envelope reductionschemes, in the nonlinearityeffect is not considered the simulations.This clearly shows that the BER are performances both the schemes more or of lessthe samewith channelcoding (Turbo coding). With the nonlinearity effect of the high of power amplifier, the BER performance the proposed scheme must improve significantly with respectto the PTS since the PAPR peris scheme much better formance the proposed of thanthe PTS.

78

Int. 2002 Thammasat J. Sc.Tech.,Vol.7, No.3, September-December

optimum algorithm for higher number of subalsorithm. channels the second in References SysP., [] Shelswell, The COFDM Modulation tem: The Heart of Digital Broadcasting, Engineering Electronicand Communication Vol 7, June,pp.127-135,1995. Journal, [2] ISO/IEC,InformationTechnology generic Coding of Moving Picturesand Associated Audio Information:Part2 video,Tech.Rep. r 3 8 1 8r S O / r E C . 2 0 0 0 . . Wu, Y. and Zou, W. Y., OrthogonalFre[3] quency Division Multiplexing: A multiIEEE Transaccarrier Modulation Scheme, Vol.4l, No. Electronics. tionson Consumer 3, August,pp 392-399,1995. OFDM Ra[4] Rohling,H., et al., Broad-band for dio Transmission MultimediaApplicaVol. tions,IEEE Proceedings., 87, October, p p . 1 7 7 8 - 1 7 81 9 9 9 . 8. [5] Jayalath,A. D. S., OFDM for Wireless Broadband Communications(Peak Power and Coding),PhD theSpectrum Reduction, VIC 3800, Clayton, sis,MonashUniversity, 2002. Australia, of andReduction Peak[6] Ochiai,H., Analysis PowerRatio in OFDM Systems, to-Average Schoolof EngiThe Graduate PhD Thesis, neering, The university of Tokyo, Japan, 2 March, 001. Muller, S. H., and Huber,J. B., OFDM with [7] Power Ratio by ReducedPeak-to-average of OptimumCombination PartialTransmisVol. 33, Letters, Electronics sion Sequence, pp. 368-369,1997. No 5, C., A. [8] Jayalath, D. S. andTellambura, Adaptive PTS Approachfor Reductionof Peakto-AveragePower Ratio of OFDM Signal, Letters,Vol. 36, No. 14, pp. Electronics 1226-1228,2000. N. [9] Cimini Jr, L. J. and Sollenberger, R., Power Ratio Reductionof Peak-to-Average an OFDM Signal Using Partial Transmit SideInformation, with Embedded Sequence ConferIEEE Global Telecommunications Vol. 2, pp.746-750,2000. ence, V., C. [10]Chong, V. andTarokh, A SimpleEnOFDM QPSK Code codableiDecodable Envelope Power with Low Peak-to-Mean Ratio. IEEE Transactionson Information

Theory, Yol. 41, No 7, pp. 3025-3029, 2001. M., et al., OFDM PeakPowerReduc[11]Tan, tion by a Novel Coding Scheme with ThresholdControl, IEEE Vehicular Technology Conference, Yol. 2, No 54ND, pp. 669-612,2001. [2]Armstrong, J., New OFDM Peak-toAverage Power Reduction Scheme,IEEE VehicularTechnology Vol. l, Conference, No 53ND,pp.756-7 2001. 60, Transform [13] Huang,X., et al., Companding for the Reduction Peak-to-Average of Power Ratio of OFDM Signals,IEEE Vehicular Vol. 2, No 53, pp. Technology Conference, 835-839,2001. N. [4] Cimini Jr, L. J. and Sollenberger, R., Peak-to-Average Power Ratio Reductionof an OFDM Signal Using Partial Transmit IEEE Communications Letters, Sequences, 2000. Vol. 4, No. 3, pp.86-88,

Parames Foomooljareon was bom 8, on in Thailand, November 1973. He receivedthe Bachelordegreein ElectricalEngineeringfiom Thammasat University, Thailand, in 1996, and Master degree in Telecommunication Engineering fiom Asian Institute of Technology,Thailand, in 2002. Since 1996 he has been with Advance Info Service positionis GSM His Pblc.Co.Ltd., Thailand. current RadioNetwork PlanningSeniorEngineer. 3 A D;| U f '. , I W.A.C. Fernando received the B.Sc. Engineerine degree (First classl in F.lectronic and TelecommunicationsEngineering fiom the Universiry of Moratuwa.

t I

fiom Asian Institute of TechTelecommunications nology, Bangkok, Thailand in 1997. He has completed his PhD at the Departmentof Electrical and Universityof Bristol, UK in ElectronicEngineering, 2001. Sincethen he has beenworking as an February professor the Asian Instituteof Technolat assistant ogy. His current researchinterestsinclude digital intelligentvideo encodimageand video processing, chaning, OFDM and CDMA for wirelesschannels, nel coding and modulation schemesfor wireless channels.He has publishedmore than 60 intemaon tional papers theseareas

"ii '",fli":li";" ::L::*'

Anda mungkin juga menyukai