Anda di halaman 1dari 5

IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 56, NO.

11, NOVEMBER 2008 1775

A Novel Selected Mapping Technique for


PAPR Reduction in OFDM Systems
Stéphane Y. Le Goff, Boon Kien Khoo, Charalampos C. Tsimenidis, and Bayan S. Sharif

Abstract—Selected mapping (SLM) is a well-known method for or the maximum likelihood decoding scheme introduced in
reducing the peak-to-average power ratio (PAPR) in orthogonal [9]. More recently, another method was proposed in [10] that
frequency-division multiplexing (OFDM) systems. The main combines channel estimation using high-power pilot tones and
drawback of this technique is that, for each data block, it requires
the transmission of several side information bits, which results PAPR reduction via SLM. One of the key ideas in [10] consists
in some data rate loss. These redundant bits are so critical to the of choosing the location of the pilot tone used inside each data
error performance of the system that they need in practice to be sub-block depending on the side information index, and then
protected by a powerful channel code. This increases the system exploit the power disparity between this pilot tone and the data
complexity and transmission delay, and decreases the data rate symbols in the same sub-block to allow for side information
even further. In this paper, we propose a novel SLM method
for which no side information needs to be sent. By considering recovery at the receiver side. The technique in [10] assumes
the example of an OFDM system using 16-QAM modulation, the use of several pilot tones in each data block, which is
it is shown that the proposed method performs very well both clearly not a realistic option in some important applications
in terms of PAPR reduction and bit error rate at the receiver such as immobile wireless communications.
output. It is however possible to suppress the practical limitations
Index Terms—Orthogonal frequency-division multiplexing, of the work in [10] by leaving out the pilot tones and replacing
peak-to-average power ratio, selected mapping, side information. them with data symbols. By doing so, we completely remove
the restrictions regarding the possible locations of the high-
power subcarriers inside the data block. We are therefore
I. I NTRODUCTION left with a problem which is much more general than that

A MAJOR drawback of orthogonal frequency-division addressed in [10]. In this paper, we propose to address this
multiplexing (OFDM) systems has traditionally been problem by describing a novel SLM method without side
their high peak-to-average power ratio (PAPR). To overcome information and studying its performance in terms of PAPR
this problem, a number of techniques have been developed reduction, probability of erroneous side information detection,
(see, e.g., [1] - [7]). Selected mapping (SLM) is one of as well as bit error rate (BER) at the receiver output. In our
the most promising among all these techniques because it is SLM method, each side information index is associated with
simple to implement (at least from a conceptual viewpoint), a particular set of locations inside the data block at which
introduces no distortion in the transmitted signal, and can the modulation symbols have been extended. In the receiver,
achieve significant PAPR reduction [2]. The basic idea in a side information detection block attempts to determine the
SLM consists of generating a set of signals, all of them locations of the extended symbols.
representing the same data block, and then transmitting the The paper is organized as follows: In Section II, the
one with the lowest PAPR. The implementation of SLM leads proposed SLM technique is presented. Then, in Section III, we
to a decrease in data rate because the selected signal index, consider an example for which we provide various computer
called side information, must also be transmitted to allow for simulation results. Finally, conclusions are drawn in Section
the recovery of the original data block at the receiver side. IV.
The eventual loss of the side information during transmission
significantly degrades the error performance of the system II. T HE P ROPOSED SLM T ECHNIQUE WITHOUT S IDE
since the whole data block is lost in this case. This is why it I NFORMATION
is, in practice, crucial to protect the side information using In this Section, a notation in the form V = (vq )Q shall be
a powerful channel code, which makes the system more used to denote a vector V composed of Q scalar quantities
complex, increases the transmission delay, and further reduces vq , q ∈ {0, 1, ..., Q − 1}.
the data rate.
It is therefore important to try implementing the SLM A. The proposed SLM transmitter
method without having to explicitly send any side information. Consider an OFDM system using N orthogonal subcarriers.
A few techniques for doing so have already been proposed A data block is a vector X = (xn )N composed of N complex
such as, for example, the scrambling method described in [8] symbols xn , each of them representing a modulation symbol
Paper approved by S. A. Jafar, the Editor for Wireless Communication The- transmitted over a subcarrier. In the classical SLM technique,
ory and CDMA of the IEEE Communications Society. Manuscript received X is multiplied element by element with U phase vectors
February 14, 2007; revised July 10, 2007. Bu = (bu,n )N composed of N complex numbers bu,n , u ∈
The authors are with the School of Electrical, Electronic and Computer
Engineering, Newcastle University, UK (e-mail: stephane.le-goff@ncl.ac.uk). {0, 1, ..., U − 1}, defined so that |bu,n | = 1, where | · | denotes
Digital Object Identifier 10.1109/TCOMM.2008.070021 the modulus operator. Each resulting vector Xu = (xu,n )N ,
0090-6778/08$25.00 
c 2008 IEEE

Authorized licensed use limited to: VELLORE INSTITUTE OF TECHNOLOGY. Downloaded on August 3, 2009 at 05:06 from IEEE Xplore. Restrictions apply.
1776 IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 56, NO. 11, NOVEMBER 2008

Sub-vector 0 Sub-vector 1 Sub-vector 2


where xu,n = bu,n ·xn , produces, after inverse discrete Fourier
transform, a corresponding OFDM signal su (t) given by u=0 C C 1 1 1 C C 1 1 1 C C 1 1 1

u=1 C 1 C 1 1 C 1 C 1 1 C 1 C 1 1
N −1  
1  2πnt u=2 C 1 1 C 1 C 1 1 C 1 C 1 1 C 1
su (t) = √ xu,n · exp j , 0≤t≤T, (1)
N n=0 T u=3 C 1 1 1 C C 1 1 1 C C 1 1 1 C

where T is the OFDM signal duration. Among the U signals u=4 1 C C 1 1 1 C C 1 1 1 C C 1 1


su (t), the one having the lowest PAPR is selected for transmis-
u=5 1 C 1 C 1 C C 1 C 1 C 1
sion. To allow for the recovery of the original vector X at the 1 1 1

receiver side, one needs to transmit log2 (U ) side information u=6 1 C 1 1 C 1 C 1 1 C 1 C 1 1 C


bits to indicate which particular vector was selected among
u=7 C C C C 1 C C
the U vectors Xu available. 1 1 1 1 1 1 1 1

In this paper, we propose a novel SLM technique that allows u=8 1 1 C 1 C 1 1 C 1 C 1 1 C 1 C


for the side information to be reliably embedded in Xu so
u=9 1 1 1 C C C C 1 1 1 C C
that no additional bits need to be sent to the receiver. In our 1 1 1
technique, the phase vectors Bu = (bu,n )N are defined so that n= 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
the terms |bu,n | are not all equal to the unit. Actually, these m= 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4
vectors are composed of some elements whose moduli can
Fig. 1. List of the 10 phase vectors Bu that can be used to apply our SLM
also be equal to a constant C > 1. technique to an OFDM system with N = 15 subcarriers, when each vector
The procedure to construct the set of phase vectors Bu = Bu is divided into L = 3 subvectors with M = 5 and K = 2. We only
(bu,n )N , u ∈ {0, 1, ..., U −1}, in the proposed SLM technique show the moduli of the elements bu,n since their phases can take any random
values.
is as follows: for a given vector Bu , the phases of the complex
elements bu,n can first be chosen randomly, as in classical
SLM. The next step consists of dividing Bu into L subvectors As an illustration, Fig. 1 shows the phase vectors Bu that
of length M = N/L. Hereafter, we denote by bm,l the (m + can be used to apply our SLM technique to an OFDM system
1)th element in the (l + 1)th subvector, m ∈ {0, 1, ..., M − 1}, with N = 15 subcarriers, when each vector Bu is divided
l ∈ {0, 1, ..., L − 1}. In each subvector, the moduli of K into L = 3 subvectors with M = 5 and K = 2. There are
elements are set to a constant C > 1, whereas the moduli of U = 52 = 10 possible vectors Bu . Note that Fig. 1 only
the other (M − K) elements remain equal to the unit. The shows the moduli of the elements bu,n since their phases can
locations in each subvector of the elements bm,l for which basically take any random values.
|bm,l | = C are identical for each subvector, i.e. the values of It is important to mention that, in the proposed SLM tech-
the terms |bm,l | do not depend on the index l ∈ {0, 1, ..., L − nique, the average energy per transmitted symbol is increased
1}. when the data block X is multiplied by the phase vectors Bu
For a given phase vector Bu , these locations form a set because the fact that |xu,n | = |bu,n | · |xn |, with |bu,n | being
Su composed of K integers. For instance, if a vector Bu either equal to 1 or C > 1, implies that E[|xu,n |2 ] > E[|xn |2 ],
is associated with the set Su = {0, 3}, it means that only where E[·] designates the expectation operator. This energy
the complex elements bu,n in positions m = 0 and m = 3 increase G, expressed in decibels (dB), is given by
in each subvector have a modulus equal to C. The set Su  
actually represents the side information that is not explicitly M + K · (C 2 − 1)
G = 10 · log10 . (2)
transmitted since it is embedded in the vector Bu . To allow for M
side information recovery at the receiver side, there must be a Such energy increase must be taken into account when as-
one-to-one correspondence between an index u and a set Su . sessing the error performance of an OFDM system using the
In other words, no two distinct vectors Bu can be associated proposed SLM method and comparing it, say, to that of an
with identical sets. The maximum number of phase vectors equivalent OFDM scheme based on classical SLM for which
Bu we can use to implement the proposed SLM method is there is no energy increase.
therefore equal to the number of distinct sets Su that can be
generated.
M  Such number is given by the binomial coefficient
B. The proposed SLM receiver
K = K!·(M−K)! .
M!

Once all the phase vectors Bu have been defined, our SLM In this paper, we assume that, for each transmitted symbol
method works exactly like the classical one, i.e. the data block xv,n , the corresponding frequency-domain sample, obtained
X is multiplied element by element with each Bu so as to after discrete Fourier transform, is given by
produce U vectors Xu = (xu,n )N , with xu,n = bu,n · xn , yv,n = hn · xv,n + nn , n ∈ {0, 1, ..., N − 1}, (3)
as well as U corresponding OFDM signals su (t). Finally, the
signal su (t) with the lowest PAPR is transmitted. Throughout where hn is a real sample representing the fading experienced
this paper, this particular signal is denoted as sv (t) and is by the nth subcarrier and nn is a complex Gaussian noise
associated with the vectors Xv = (xv,n )N and Bv = (xv,n )N . sample with zero-mean and variance σ 2 . Equation (3) implies
The index v ∈ {0, 1, ..., U −1} represents the side information that each subcarrier experiences a flat fading channel, which is
to be transmitted. a realistic assumption for most practical OFDM systems. Note

Authorized licensed use limited to: VELLORE INSTITUTE OF TECHNOLOGY. Downloaded on August 3, 2009 at 05:06 from IEEE Xplore. Restrictions apply.
LE GOFF et al.: A NOVEL SELECTED MAPPING TECHNIQUE FOR PAPR REDUCTION IN OFDM SYSTEMS 1777

that, throughout this work, we assume for simplicity sake that N = 125, 8 dB N = 255, 8 dB
N = 510, 8 dB N = 1020, 8 dB
all fading and noise samples are independent, although this N = 125, 25 dB N = 255, 25 dB
may not always be the case in some OFDM systems [11]. Fi- N = 510, 25 dB N = 1020, 25 dB
nally, we consider that the fading samples are perfectly known 1.E +00

at the receiver side, i.e. perfect channel state information (CSI)


is assumed.
The receiver must recover the side information index v by 1.E -01
processing both vectors Yv = (yv,n )N and H = (hn )N . This
receiver has the knowledge of the system parameters M , L,
and K as well as the one-to-one correspondence between an

Pdf
1.E -02
index u and a set Su . In other words, it knows that each
M -sample subvector inside Yv corresponds to a transmitted
subvector composed of K symbols that were extended by a
factor C and (M − K) symbols that were not extended. For 1.E -03

this receiver, determining the K locations associated with the


extended symbols in a subvector, i.e. the set Sv , is equivalent
to detecting the side information index v. 1.E -04
The detection algorithm consists of first dividing Yv into L 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9
Constant C
subvectors of length M , and then finding the K positions in
which the average energy per symbol was increased in each Fig. 2. Probability of detection failure (Pdf ) of the side information with
subvector. Let us denote by ym,l and hm,l the received and the proposed algorithm as a function of the constant C, for SNR = 8 dB and
fading samples, respectively, in (m + 1)th position in the (l + 25 dB and N = 125, 255, 510, and 1020 subcarriers. For each subcarrier, we
consider the uncoded transmission of 16-QAM symbols over a flat Rayleigh
1)th subvector, m ∈ {0, 1, ..., M − 1}, l ∈ {0, 1, ..., L − 1}. fading channel with perfect CSI. The length of the subvectors is M = 5
We can show that the average energy per symbol, Es,m , in symbols and K = 2 symbols are extended in each subvector.
position m in a subvector is given by
III. E XAMPLE
E[|ym,l |2 ] − σ 2
Es,m = . (4) To illustrate the proposed SLM technique, we consider in
E[(hm,l )2 ]
this Section the example of an OFDM system using 16-QAM
Hence, the computation of (4) for each position m allows us to over N subcarriers. The length of the subvectors is M = 5
determine the K positions with the highest energy per symbol. symbols and K = 2 symbols are extended in each subvector.
The evaluation of the expectation terms in (4) is performed by With such parameters, PAPR reduction can thus be achieved
taking into account the L samples available for each position 
using U = 52 = 10 possible phase vectors Bu . Fig. 1 shows
m, i.e. using the list of these vectors Bu . The phases of the elements bu,n ,
u ∈ {0, 1, ..., U − 1}, n ∈ {0, 1, ..., N − 1}, are randomly set
1 
L−1
E[|ym,l |2 ] ≈ · |ym,l |2 (5) to either 0 or π (with equal probabilities). All results given
L
l=0 below are obtained by computer simulations assuming that the
and fading samples follow a Rayleigh distribution defined so that
E[(hm,l )2 ] = 1.
1 
L−1
E[(hm,l )2 ] ≈ · (hm,l )2 , m ∈ {0, 1, ..., M − 1}. (6)
L A. Probability of side information detection failure
l=0

If L is not large enough, (5) and (6) may yield inaccurate Fig. 2 shows the probability of detection failure, Pdf , as a
evaluations of the expectation terms. The computation using function of the constant C, for two values of the signal-to-
(4) of the M average energies Es,m based on such evaluations noise ratio (SNR = 8 and 25 dB) and four different numbers
can then lead to erroneous detection of the set Sv , i.e. the index of subcarriers (N = 125, 255, 510, and 1020). Note that
v. In such case, the whole original vector X is lost since the the numbers of subcarriers are chosen to be multiples of
receiver is not able to recover the side information. M = 5 close to the usual powers of two used in practice.
In order to minimize the probability of side information The parameter Pdf represents the probability that the receiver
detection failure, Pdf , it is necessary to maximize the value of cannot recover the side information, i.e. a complete OFDM
L for a given number N of subcarriers, which is equivalent frame (vector X) is lost. The SNR is defined as the ratio
to minimizing the length M of a subvector because M = between the average energy per transmitted bit, Eb , and the
N/L. For a desired number U of phase vectors, the value of one-sided power spectral density, N0 , of white Gaussian noise.
M must however remain large enough so that the condition From Fig. 2, it is observed that the SNR does not have
U≤ M K is satisfied. This can be achieved by taking K = 2
M
a great influence on the value of Pdf . The latter is, on the
for even values of M and K = 2 or K = 2 for odd
M−1 M+1
other hand, much more dependent on the values of parameters
values of M . In any case, the system designer must keep in C and N . As C is increased, the performance of our algo-
mind that, for a constant M , a higher value of K results in rithm improves, simply because a higher value of C allows
a higher energy increase G, which in turns degrades the error for a better distinction between extended and non-extended
performance of the system. symbols after transmission through the channel. Increasing N

Authorized licensed use limited to: VELLORE INSTITUTE OF TECHNOLOGY. Downloaded on August 3, 2009 at 05:06 from IEEE Xplore. Restrictions apply.
1778 IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 56, NO. 11, NOVEMBER 2008

1.E+00 N = 65 (class. SL M ) N = 65 (new SLM)


N = 65 (no PAPR reduct.) N = 125 (class. SLM)
N = 125 (new SL M ) N = 125 (no PAPR re duct.)
N = 255 (class. SLM) N = 255 (new SL M)
1.E-01 N = 255 (no PAPR reduct.) N = 510 (class. SLM)
N = 510 (new SL M ) N = 510 (no PAPR reduct.)
1.E +00
1.E-02
BER

Perfect SI
1.E-03 1.E -01
C = 1.1

Pr[PA PR >PAPR0]
C = 1.2

1.E-04
C = 1.3
C = 1.4 1.E -02
C = 1.5
1.E-05
0 5 10 15 20 25 30 35 40
Eb/N0 (dB) 1.E -03

Fig. 3. BER performance of an OFDM system using 16-QAM over N = 255


subcarriers, for different values of the constant C. For each subcarrier, we
assume transmission over a flat Rayleigh fading channel with perfect CSI. The
length of the subvectors is M = 5 symbols and K = 2 symbols are extended 1.E -04
in each subvector. For comparison purposes, the BER curve obtained with the 5 6 7 8 9 10 11 12
classical SLM method with perfect side information (SI) is also plotted. PAPR 0 (dB )

also results in a better performance because the number of Fig. 4. CCDF of the PAPR of the proposed SLM technique (new SLM)
without side information for C = 1.4 and N = 65, 125, 255, and 510
subvectors L = N/M is then also increased, which provides subcarriers. The length of the subvectors is M = 5 symbols and K = 2
more reliable estimates of the average energies per symbol symbols are extended in each subvector. For comparison purposes, we also
Es,m in positions m ∈ {0, 1, ..., M − 1}. plot for each value of N the curves obtained with the classical SLM method
(class. SLM) as well as those associated with an OFDM system without any
PAPR reduction (no PAPR reduct.).

B. Bit error rate performance


It is also important to study the error performance degra- Fig. 3 confirms the presence of an error floor in the BER
dation caused by the application of the technique proposed in curves. As expected, the error floor can be lowered by increas-
this paper. Such degradation is due to both the energy increase ing the value of C, i.e. reducing the probability of detection
G and the occasional side information detection failure events. failure. Fig. 3 also shows that, if C ≥ 1.4, the error floor
In fact, for the example considered in this Section, the bit error appears only at very high SNRs (> 30 dB), which indicates
probability Peb at the receiver output is given by that it may not be a major issue for many practical OFDM
systems that tend to operate at lower SNRs. For instance, if
Pdf
Peb = P · (1 − Pdf ) + , (7) we target a BER of 10−2 at the 16-QAM demodulator output,
4 as may well be the case in coded OFDM systems, then the
where P designates the bit error probability obtained when the application of our technique with C = 1.4 or 1.5 degrades
side information is properly detected. As Eb /N0 → +∞, P → the error performance by only 0.5 dB when compared to the
P
0, and thus Peb → 4df , which shows that the erroneous side classical SLM method with perfect side information.
information detection events introduce an error floor in the
error probability curve. It is crucial to minimize Pdf to make C. PAPR reduction performance
sure that this error floor is as low as possible. However, (7) also
indicates that the side information detection failure events can Finally, Fig. 4 shows the complementary cumulative dis-
actually be ignored if the OFDM system operates at low SNRs tribution function (CCDF) of the PAPR obtained with the
P
for which we have P  4df . For practical OFDM schemes proposed SLM technique, for C = 1.4 and N = 65, 125, 255,
operating at such SNRs (typically, coded OFDM systems), and 510 subcarriers. The curves corresponding to the ordinary
failing to properly detect the side information may therefore OFDM without PAPR reduction and the classical SLM method
not be a major issue as long as it does not happen too often. are also depicted for comparison purposes. These results are
obtained by using an oversampling factor equal to 4 [12]
In Fig. 3, we have plotted the bit error rate (BER) versus
and take into account the average energy increase introduced
Eb /N0 curve for the OFDM system considered in this Section
by our method. We observe that, for all configurations, the
with N = 255 subcarriers, for several values of the constant
performance of our SLM method in terms of PAPR reduction
C. These results take into account the increase G, given by
is identical to that of the classical SLM technique.
(2), in average energy per transmitted symbol introduced by
our algorithm. For comparison purposes, we have also plotted
the BER curve obtained with an equivalent OFDM system IV. C ONCLUSION
using the classical SLM technique with perfect, i.e. error-free, We have proposed a simple SLM technique that does not
side information. require the explicit transmission of side information. A study

Authorized licensed use limited to: VELLORE INSTITUTE OF TECHNOLOGY. Downloaded on August 3, 2009 at 05:06 from IEEE Xplore. Restrictions apply.
LE GOFF et al.: A NOVEL SELECTED MAPPING TECHNIQUE FOR PAPR REDUCTION IN OFDM SYSTEMS 1779

of our method has been carried out by considering the use [2] R. W. Bäuml, R. F. H. Fischer, and J. B. Huber, “Reducing the peak-
of 16-QAM over each subcarrier. When compared to the to-average power ratio of multicarrier modulation by selected mapping,”
Electron. Lett., vol. 32, no. 22, pp. 2056-2057, Oct. 1996.
classical SLM method using error-free side information, we [3] X. Li and L. J. Cimini, Jr., “Effect of clipping and filtering on the
found that the only price to pay for the implementation of our performance of OFDM,” IEEE Commun. Lett., vol. 2, no. 5, pp. 131-
method is a slight degradation in error performance as well 133, May 1998.
[4] H. Ochiai and H. Imai, “Performance of the deliberate clipping with
as a small complexity increase at the receiver side (presence adaptive symbol selection for strictly band-limited OFDM systems,” IEEE
of a side information detection block). We showed that the J. Select. Areas Commun., vol. 18, no. 11, pp. 2270-2277, Nov. 2000.
probability of erroneous detection of the side information can [5] H. Ochiai and H. Imai, “Performance analysis of deliberately clipped
OFDM signals,” IEEE Trans. Commun., vol. 50, no. 1, pp. 89-101, Jan.
actually be made very small in a realistic OFDM transmission 2002.
environment. For a given set of system parameters (channel [6] B. S. Krongold and D. L. Jones, “PAR reduction in OFDM via active
type, number of subcarriers, length of subvectors, etc.), we constellation extension,” IEEE Trans. Broadcast., vol. 49, no. 3, pp. 258-
268, Sept. 2003.
believe that this probability will be even smaller if 16-QAM [7] S. H. Han, J. M. Cioffi, J. H. Lee, “Tone injection with hexagonal
is replaced by QPSK but will worsen in case 64-QAM or constellation for peak-to-average power ratio reduction in OFDM,” IEEE
256-QAM is used in place of 16-QAM. The application of Commun. Lett., vol. 10, no. 9, pp. 646-648, Sept. 2006.
[8] M. Breiling, S. H. Müller-Weinfurtner, and J. B. Huber “SLM peak-power
our method to higher-order signal constellations is thus an reduction without explicit side information,” IEEE Commun. Lett., vol.
issue that needs to be investigated. 5, no. 6, pp. 239-241, June 2001.
Throughout this work, we have assumed that the received [9] A. D. S. Jayalath and C. Tellambura, “SLM and PTS peak-power
reduction of OFDM signals without side information,” IEEE Trans.
samples yv,n within one OFDM frame are independent. In Wireless Commun., vol. 4, no. 5, pp. 2006-2013, Sept. 2005.
future studies, it may also be interesting to assess the per- [10] N. Chen and G. T. Zhou, “Peak-to-average power ratio reduction in
formance of the proposed SLM method by assuming some OFDM with blind selected pilot tone modulation,” IEEE Trans. Wireless
Commun., vol. 5, no. 8, pp. 2210-2216, Aug. 2006.
correlation between the received samples yv,n . [11] M. Russell and G. L. Stüber, “Interchannel interference analysis of
OFDM in a mobile environment,” in Proc. IEEE Vehicular Technology
R EFERENCES Conf., vol. 2, Chicago, IL, July 1996, pp. 820-824.
[12] C. Tellambura, “Computation of the continuous-time PAR of an OFDM
[1] A. E. Jones, T. A. Wilkinson, and S. K. Barton, “Block coding scheme signal with BPSK subcarriers,” IEEE Commun. Lett., vol. 5, no. 5, pp.
for reduction of peak to mean envelope power ratio of multicarrier 185-187, May 2001.
transmission scheme,” Electron. Lett., vol. 30, no. 25, pp. 2098-2099,
Dec. 1994.

Authorized licensed use limited to: VELLORE INSTITUTE OF TECHNOLOGY. Downloaded on August 3, 2009 at 05:06 from IEEE Xplore. Restrictions apply.