Sie sind auf Seite 1von 6

Limits of Phase Noise Suppression in OFDM

Denis Petrovic, Wolfgang Rave and Gerhard Fettweis


Vodafone Chair Mobile Communications Systems, Dresden University of Technology, Dresden, Germany.
e-mail: {petrovic, rave, fettweis}@ifn.et.tu-dresden.de.

Abstract: We introduce two independent approaches for burst like errors, resulting in the performance error floor.
phase noise suppression. The dominant effects, responsible Consideration of the system capacity in the presence of
for the degradation of an OFDM system performance, if the phase noise in Sec. 3., motivates the idea, that bit
phase noise is present, are identified. We found out, that interleaving can significantly improve the system perfor-
the system performance is strongly influenced by certain mance. For systems where interleaving delay is a critical
phase noise realizations, which cause burst errors, result-
issue, we propose an iterative algorithm for phase noise
ing in the performance error floor. Consideration of the
suppression. We presented the idea to iteratively sup-
system capacity in the presence of phase noise, motivates
the idea, that bit interleaving can significantly improve the press the phase noise in [16]. In Sec. 4. one realization
system performance. For systems where interleaving de- of this idea is presented and in Section 5. the algorithm
lay is a critical issue, we propose, as a second approach, an performance is verified by numerical simulations.
iterative phase noise suppression algorithm. Simulation re-
sults in terms of packet error rate (PER) show, that both 2. System Model
bit interleaving and iterative algorithm are capable to sig- Consider an OFDM transmission over N subcarriers,
nificantly suppress the phase noise.
as shown in Fig. 1. For simplicity, we assume the direct
1. Introduction conversion approach where both up- and downconver-
sion are done in one step [8].
OFDM has been applied in a variety of digital com- In the case of perfect frequency and timing syn-
munications applications due to its robustness to fre- chronization the received OFDM signal samples in
quency selective fading. However, OFDM is very sen- the presence of phase noise can be expressed as
sitive to synchronization errors, one of them being phase r(n) = (x(n) ⋆ h(n))ejφ(n) + ξ(n). The variables
noise [4]. Phase noise reflects imperfections of the local x(n), h(n) and φ(n) denote the samples of the transmit-
oscillator (LO), i.e. random drift of the LO phase from ted signal, the channel impulse response and the phase
its reference. noise process at the output of the mixer, respectively.
There are two effects that occur if the phase noise is The symbol ⋆ stands for convolution. The term ξ(n) rep-
present in an OFDM system [1]: rotation of all demod- resents AWGN noise.
ulated subcarriers of an OFDM symbol by a common The phase noise process φ(t) is modelled as a Wiener
angle, called common phase error (CPE) and the occur- process [5] [12], with a certain 3dB bandwidth ∆f3dB .
rence of the intercarrier interference (ICI). The CPE re- To characterize the quality of an oscillator in an OFDM
sults from the DC value of the phase noise and the ICI system the relative phase noise bandwidth δP N =
comes from the deviations of the phase noise from its ∆f3dB /∆fcar is used, where the ∆fcar is the subcarrier
DC value, during one OFDM symbol. spacing. The reason for this is that δP N parameter incor-
The problem of suppressing phase noise in OFDM porates both the phase noise and system parameters.
systems can be understood as getting as much informa- Since we use a discrete time model, we need the dis-
tion on the phase noise waveform as possible. Once one crete time model of the phase noise. The discrete time
has this information, it can be used to remove the ef- equation for the Wiener phase noise process can be writ-
fects the phase noise. The simplest approach would be ten as [6] [5]:
to approximate the phase noise with a constant value,
i.e. its mean [13, 17, 19]. More advanced approaches φ(n + 1) = φ(n) + w(n) (1)
try to estimate higher spectral components to get better
approximation of phase noise waveform, thus reducing where φ(n) denotes the phase noise process at sampling
ICI [3, 11, 20]. Suppressing ICI is of large importance, instant nTs at the receiver, n ∈ Z and w(n) is a Gaussian
especially if bandwidth efficient higher order modula- random variable w(n) ∼ N (0, 4π∆f3dB Ts ).
tions need to be employed or if the spacing between the The result after the discrete Fourier transform (DFT)
carriers is to be reduced. at the receiver can be obtained by the following reason-
In this paper we concentrate only on the effects of ICI ing. Phase noise affects the received signal as an angular
on OFDM transmission. CPE is already corrected for in multiplicative distortion. Multiplication of two signals in
wireless standards using pilots [7]. the time domain is equivalent to convolving the spectra
Understanding ICI is a very important issue, because of the corresponding signals in the frequency domain.
it can lead researchers to new algorithms for its suppres- To be precise, since the discrete signals are considered
sion. Here we focus on identifying the dominant effects, here, in the frequency domain (discrete fourier transform
that are responsible for the performance degradation of domain) the spectra of two signals are circularly con-
an OFDM system, if the phase noise is present. We have volved [9]. Therefore, at the receiver, after removing the
found out, that the system performance is strongly influ- cyclic prefix and taking the DFT on the remaining sam-
enced by certain phase noise realizations, which cause ples, the demodulated carrier amplitudes Rm,s at subcar-
OFDM Modulator Upconversion demodulator block at the receiver, i.e. OFDM modula-
X m ,s s = 0,1, 2...N -1
IFFT CP LPF tion/demodulation, channel effects and phase noise. Ca-
x(t )
pacity is chosen as the information theoretic parameter,
x(n) = x(nTs )
e j 2π f ct to describe the system performance. Capacity is the ul-
Channel
− ( j 2π f c t −φ ( t ))
fs
e timate limit for the data rate which can be achieved in
Rm ,s s = 0,1, 2...N -1 a system. Consider the memoryless discrete input and
FFT CP
continuous output channel, with input x from alphabet
r (n) = r (nTs ) r (t ) = [ x(t ) ∗ h(t ) ] e jφ (t )
X , output y and transition distribution pY |X (y|x). Then
OFDM Demodulator Downconversion
the capacity under uniform inputs constraints and perfect
channel state information is given by [2]:
Figure 1: OFDM Transmission in the presence of phase
" P #
noise. pY |X (y|x)
x∈X
C = b − Ex,y log2 (4)
pY |X (y|x)
th
rier s (s = 0, 1, ...N − 1) of the m OFDM symbol are
given as: where b is the number of bits, transmitted over QAM
Rm,s = Xm,s Hm,s Im (0) + symbol, which is mapped to one subcarrier. Increasing
| {z } b, increases maximum throughput of the system. New
CP E
wireless standards consider using more and more signal
N −1
X points in a signal constellation.
Xm,v Hm,v Im (s − v) +ηm,s (2)
v=0
Assuming that BICM is used [2], we have determined
v6=s
| {z } the capacity of an OFDM transmission over an AWGN
ICI channel with added phase noise. For calculating the ca-
pacity, the equivalent channel transition (pdf) pY |X (y|x)
where Xm,s , Hm,s and ηm,s represent transmitted sym-
is required. We resorted to a Monte Carlo method to
bols on the subcarriers, the sampled channel transfer
obtain this pdf and we have numerically calculated the
function at subcarrier frequencies and transformed white
capacity. Since we are interested only in the effects of
noise which remains AWGN. The terms Im (i), i =
ICI on OFDM transmission, we assume that CPE is ide-
−N/2, ..., N/2 − 1 correspond to the DFT of the real-
ally corrected for. In Fig. 3 the capacity for different M-
ization of ejφ(n) during one OFDM symbol:
QAM OFDM signaling schemes over an AWGN channel
N −1
1 X −j2πni/N jφ(n) as a function of the relative phase noise bandwidth δP N
Im (i) = e e (3) is plotted (the corresponding curves are denoted as: with
N n=0
real phase noise). This figure can be assumed as a limit
In Eq. (2) the multiplicative distortion term Im (0) com- since the SN R for which it is simulated is very large,
mon to all subcarriers of one OFDM symbol, corre- namely SN R = 30 dB.
sponds to the common phase error (CPE). The CPE From the capacity curves we see that the phase
equals the DC value of the phase noise and must be cor- noise has a much stronger influence on the capacity
rected for to obtain acceptable performance. The inter- of higher order constellations, as this dependence gets
carrier interference (ICI) part is the additional error term much steeper near the origin with an increase of the sig-
caused by non-zero frequency components of the phase nal points. Even for very small δP N it is impossible to
noise process. It is a mixture of channel transfer function transmit 8 bits over the channel.
coefficients, transmitted symbols and phase noise terms. Further, we compare this benchmark system with
It is found that the ICI term is non-Gaussian distributed ”real” phase noise and ideal CPE correction, with the
2 system, where the ICI term after DFT is replaced with a
random variable [10] [14] [15] of power σICI . Inter-
2 2
carrier interference power σICI can be calculated in the Gaussian random variable of variance σICI before the
closed form, using several different approaches [11] [14] demodulator. Note that in the latter case we use the
[15] [18]. standard formula for the calculation of the capacity over
gaussian channels.
3. Capacity of an OFDM System with Fig. 3 shows that the performance of the system with
Phase Noise ”real” phase noise is different from that, where ICI is as-
An equivalent representation of the coded OFDM sumed to be gaussian distributed. This is to be expected
transmission system, according to the IEEE802.11a considering the discussions in [10] [15] [14] which point
standard [7], is shown in Fig. 2. The building blocks of out that the ICI is non-gaussian distributed. Indeed, if the
this scheme are encoder, interleaver (π), symbol map- ”real” phase noise is present, the capacity curves start to
per, channel with transition probability density func- fall off earlier than in the case where ICI is assumed to
tion (pdf) pY |X (y|x), demodulator (branch metric com- be gaussian. However, one should notice that the differ-
puter), branch metric deinterleaver (π −1 ) and decoder. ences of the capacity curves are not large, which moti-
Encoder, interleaver and mapper are the building blocks vates the discussion of the next section.
of the bit interleaved coded modulation (BICM) [2]. In 3.1. Phase noise ”suppression” using Interleaving
our case pY |X (y|x) describes all effects of an equiva- In [14] the distribution of ICI and its influence on the
lent ”channel” between mapper at the transmitter and the symbol error rate has been investigated. It was con-
Encoder π Mapper pY|X(y|x) Demodulator π -1 Decoder

Figure 2: Equivalent OFDM transmission block diagram.

0
8 10
with real phase 64 QAM OFDM
7 noise AWGN Channel
256 QAM OFDM ICI gaussian
C [bits/transmission]

approximation -1
δ PN=2 ⋅ 10-3
6 10
64 QAM OFDM SNR=30dB

PER
5 no phase noise

-2
ICI modelled as 1
4 10 Gaussian r.v. and
16 QAM OFDM variance σ 2ICI
3
5
2 -4 10
-3 10
-3 -2 -1
10 10 10 10 0 5 10 15 20 25
δ PN E b/N0 [dB]

Figure 3: Phase noise influence on system performance: Figure 4: Effect of the interleaving on the OFDM trans-
Capacity of the system with phase noise is compared to mission in the presence of phase noise for AWGN chan-
the capacity of the system where ICI is assumed to be nels
gaussian distributed
per correspond to the IEEE802.11a standard. We use
cluded that the distribution of ICI has significant influ- 64QAM modulation, standard convolutional code with
ence on the symbol error rate (SER) of the system. rate r = 1/2 and random interleaving. One transmis-
sion block consists of 10 OFDM symbols which com-
The statistics of the phase noise characterize this ran-
prise one code word. Number of 10000 packets is trans-
dom process on the long term. It would be interesting
mitted, to assure valid statistics. Hard decision Viterbi
at this point to concentrate on the specific realizations
decoder is used at the receiver.
of the process, i.e. realizations of the phase noise dur-
Fig. 4 shows the packet error rate (PER) dependence
ing one OFDM symbol and relate it with ICI, within one
on the interleaving depth, for coded OFDM transmis-
OFDM symbol interval.
sion over AWGN channels and phase noise with δP N =
If the phase noise does not change within one OFDM 2 · 10−3 . The interleaving depth is varied, i.e. it amounts
symbol, then the ICI term is zero. The more the phase 1, 5 and 10 OFDM symbols, within one code word. The
noise changes the larger is the ICI. Normally phase noise performance of the system changes drastically as the in-
oscillates around zero. However, within some OFDM terleaving depth varies. The reason for that is that the
symbols the phase noise can change dramatically (al- bad events, i.e. symbols which suffer from large ICI, are
most linearly) in one direction and in this case ICI is spread over a code word and the decoder tackles better
very large. These events spread the distribution of ICI, with these events. The limit is the curve, where ICI is
producing tails, which are much more pronounced than assumed to be gaussian with variance σICI 2
. For AWGN
by Gaussian distribution [14]. The occurrence of such channels without phase noise, and for the case, where
phase noise realizations cause burst errors, which de- ICI is assumed Gaussian, interleaving depth does not
grade the performance of the system. Coming back to play any role.
the discussion on the capacity of an OFDM system in the Thus one approach for phase noise ”suppression” is to
presence of phase noise, we note, that the expression for increase the interleaving depth. Capacity curves justify
capacity evaluation Eq. (4) assumes ideal (infinite) inter- this approach as they show that OFDM transmission is
leaving. Therefore the capacity curves give us the result, not much sensitive to phase noise. Thus bit interleaving
which seems to be reasonable, namely: If one uses ideal can be considered as a means for improving performance
interleaving, then the capacity of the system with phase if the delay due to interleaving is not critical and the code
noise is close to the capacity, where ICI is assumed to be words are not to short.
gaussian distributed, and this means larger capacity. It Even though, this sounds to be reasonable approach
follows, that one simple way to improve the performance to easily suppress the phase noise, to our knowledge it
of the OFDM system in the presence of phase noise is to has not been considered in the literature as a means for
increase the interleaving depth. the phase noise suppression. If the interleaving depth
3.2. Numerical Experiment cannot be changed, then the phase noise suppression in
We have examined this expectation by using computer standard systems is limited to a CPE correction, which
simulations. System parameters used throughout this pa- is estimated using pilots.
In the next section, we review an approach for phase bols. As transmitted symbols estimates, the symbols af-
noise suppression we presented in [11], and present ter necessary CPE correction are adopted. Once the DFT
a new iterative algorithm for phase noise suppression coefficients of the phase noise are known, one possesses
based on this approach. enough information on the phase noise waveform, in or-
der to suppress it.
4. Phase Noise Approximation and Cor- The phase noise suppression in the time domain would
rection be a logical approach. One should multiply the received
4.1. ICI Correction - Idea signal r(n) = (x(n) ⋆ h(n))ejφ(n) + ξ(n) with the es-
timate of e−jφ(n) . Multiplication in the time domain for
A phase noise compensation beyond the simple CPE
discrete time systems is mapped to the circular convolu-
correction will be possible only if one knows instanta-
tion of DFT spectra in the frequency domain [9]. This
neous realization of the phase noise process. The al-
means that the ICI cancellation for the mth OFDM sym-
ready introduced factors Im (i), i = −N/2, ..., N/2 − 1
bol can be done in the frequency domain by circularly
(see Eq. (3)) represent the DFT coefficients (spectral
convolving the demodulated symbols vector of all sub-
components) of one realization of the random process
carriers Rm,N = [Rm (0), . . . , Rm (N − 1)]T with the
ejφ(n) . The more spectral components Im (i) of the sig-
vector of estimated DFT coefficients of e−jφ(n) .
nal are known, the more is known of the signal wave-
The concrete realization of an algorithm is as follows:
form ejφ(n) , and thus φ(n). The signal ejφ(n) has the
characteristics of a low-pass signal [5], with power spec- 1. Step 1: Perform standard CPE correction using least
tral density of the form 1/(1 + f 2 ), where f denotes the square (LS) estimation [19] [17].
frequency. Additionally, phase noise has a very small
2. Step 2: Make a decisions on the transmitted sym-
bandwidth compared with the subcarrier spacing. Due
bols and use all available hard decisions for the
to the shape of the spectrum of ejφ(n) , very few low pass
MMSE estimation of the I˜m (i), i = −u...u, ac-
spectral components will suffice to give a ”good” ap-
cording to the method provided in [14]. Here u de-
proximation of the phase noise waveform. This is illus-
notes the order of the phase noise approximation.
trated by the example in Fig. 5, where it can be seen that
already second order approximation gives much better 3. Step 3: Convolve the vector of the received symbols
phase noise approximation than only DC value. There- with the DFT coefficients of the conjugate of the
fore knowledge of the coefficients Im (i) gives the pos- phase noise waveform.
sibility to approximate the phase noise waveform to a
4.3. Iterative Phase Noise Suppression
higher order, and allows a better compensation of it than
with only CPE correction, i.e. 0th order approximation. Described algorithm for ICI suppression is the deci-
sion feedback algorithm. It is to expect that falsely de-
tected symbols after initial CPE correction, which are
fed to the MMSE estimator in the Step 2 of the algo-
0.14 Phase noise trajectory (δ
PN
=1⋅ 10-3)
rithm, influence the estimation process. The reduction
0-th Order approx. (CPE) of the symbol error of the symbols, which are fed back,
0.12 Im(0)
will improve the quality of the phase noise estimation
angle [rad]

and thus the quality of the phase noise suppression. This


0.1
can be achieved if the algorithm described in Sec. 4.2. is
applied iteratively.
0.08 1st Order approx.
Im(0), Im(1), Im(-1) As discussed in Sec. 4.1. it is to expect that the non-
iterative ICI suppression algorithm gives better perfor-
0.06 2nd Order approx.
Im(0), Im(1), Im(-1),Im(2), Im(-2) mance, i.e. reduced symbol error rate, than the pure CPE
correction. If this ”better” symbols are used, to again es-
0.04
10 20 30 40 50 60 timate the phase noise, it is to expect that the system
samples performance will be improved.
The proposed algorithms is realized in three steps:
Figure 5: Phase noise waveform approximation using
1. Step 1: Perform standard CPE correction using least
various orders of approximation.
square (LS) estimation [19] [17].
2. Step 2: Carry out ICI suppression algorithm de-
4.2. ICI Correction Algorithm scribed in the previous section (only Steps 2 and
The details of an ICI suppression algorithm can be 3),
found in our previous work [11] [14]. Proposed ICI sup-
3. Step 3: Demodulate QAM symbols and feed them
pression algorithm estimates as many spectral compo-
back to the Step 2. Iterate until the desired perfor-
nents Im (i),i = −N/2, ..., N/2 − 1 as possible using
mance achieved.
MMSE estimation. The information about these spectral
components is hidden in the ICI part of the signal at the A block diagram of this scheme is presented in Fig. 6.
output of the DFT demodulator Rm,s , s = 0, 1, ...N − 1 The complexity of the algorithm is large, however the
given by Eq. (2). Estimation algorithm is a decision performance of the algorithm is dramatically improved
feedback algorithm, since it requires transmitted sym- as will be shown in Sec. 5..
0
10
Demodulator Dem
bits 64 QAM
Estimate
Im(i)
Deconv Viterbi iterate
δ PN = 5⋅ 10-3 ICPE LS
I m ( i ) , i = −u ,..., u
yes
Rm ,lk lk = 0,1, 2...N -1 ETSI A Channel
-1
10

PER
Xˆ m,lk lk = 0,1, 2...N -1
CPE Correction
3
ICI supp. alg
-2
10 genie ICI supp. 1-iter
Step 1
3-iter
Step 2 Step 3

feedback no PN 3
Reconstruct
Tx symbols -3
10
Xˆ m′ ,lk lk = 0,1, 2...N -1 0 10 20 30 40
E b/N0

Figure 6: Block diagram of the iterative phase noise sup-


Figure 7: Performance of the iterative ICI suppression
pression algorithm.
algorithm for ETSI A channel and δP N = 5 · 10−3 .

0
10
5. Numerical Results 64 QAM -3
δ PN = 5⋅ 10
System parameters correspond to that described in ICPE
AWGN Channel
Sec. 3.2.. Within simulations six scenarios are com- 10
-1 LS
pared: 1) without phase noise (no PN); 2) with phase
no PN
PER

noise and genie CPE correction (ICPE); 3) with phase 3 iter


noise and CPE correction using least squares (LS) algo- ICI supp. alg
-2
rithm [19]; 4) with phase noise and genie ICI correction 10
genie ICI supp.
of certain order u; 5) with ICI correction of uth order and
6) with phase noise and an iterative phase noise suppres- 1 iter 3
sion (number of iterations denoted by numbers). 3
-3
10
First set of simulation results in terms of PER is plot- 0 5 10 15 20 25
E b/N0
ted in Figs. 7 and 8 for the ETSI A channel and an
AWGN channel, respectively. Adopted relative phase
noise bandwidth is δP N = 5 · 10−3 . The ICI correc- Figure 8: Performance of the iterative ICI suppression
tion order adopted is u = 3. The ICI correction algo- algorithm for AWGN channel and δP N = 5 · 10−3 .
rithm shows better performance than the pure CPE cor-
rection, however, the results are much worse than the
achievable genie correction of the specified order. This can be achieved.
performance limitation is due to the decision feedback The performance presented here is in terms of the
nature of the algorithm. Falsely detected symbols, from packet error rate (PER). It is interesting to note that the
Step 2 of an algorithm, which are used for estimation of bit error rate performance can even worsen with increas-
the phase noise DFT coefficients, will influence the es- ing number of iterations, while the PER decreases. For
timation process. This problem is more pronounced if OFDM symbols, for which, after the initial CPE correc-
the phase noise bandwidth is large, because then ICI is tion, many subcarriers are erroneously detected, the ICI
large, which influences also the estimation of the CPE. estimation can produce additional errors. In the iterative
To improve the performance of this algorithm, or in algorithm this causes error propagation. However, for
other words to reduce the error propagation problem, the OFDM symbols with only few falsely detected subcarri-
iterative approach for phase noise suppression should be ers, ICI algorithm is capable of correcting these errors.
considered. This algorithm provides results that are very Packets with few errors will be recovered by the algo-
close to genie phase noise suppression of the correspond- rithm, while packets with many errors after the initial
ing order (see Figs. 7 and 8). However the complexity of CPE correction will probably have even more errors.
the algorithm is quite large. Therefore, the quality of 6. Conclusions
the phase noise suppression is the trade off between the
complexity and performance. Fig. 9 shows an addi- In this paper we introduced two new approaches for
tional example for the same scenarios, ETSI A channel phase noise suppression in OFDM. It was shown that the
and δP N = 2 · 10−3 . The conclusions are similar as bit interleaving is a very simple, still effective means of
for the δP N = 5 · 10−3 . The ICI correction of order suppressing the effects of the phase noise. For the appli-
u = 7 has been performed. Simulation results show cations where the interleaving depth cannot be large, e.g.
that for both frequency selective and AWGN channels, for short code words, we propose an iterative algorithm
and both large and small phase noise bandwidths, per- for phase noise suppression. The performance of the al-
formance very close to genie correction of certain order gorithm is very close to the genie suppression of certain
10
0 [9] A.V. Oppenheim and R.W. Schafer. Discrete-Time
Signal Processing. Prentice-Hall Inc., 1989.
64 QAM -3
δ PN = 2⋅ 10 [10] D. Petrovic, W. Rave, and G. Fettweis. Phase Noise
-1
ETSI A Channel Influence on Bit Error Rate, Cut-off Rate and Ca-
10
no PN pacity of M-QAM OFDM Signaling. In Proc. Intl.
ICPE
PER

OFDM Workshop (InOWo)02, 2002.


ICI supp. alg LS
-2 [11] D. Petrovic, W. Rave, and G. Fettweis. Phase
10 genie ICI supp. Noise Suppression in OFDM including Intercarrier
1-iter Interference. In Proc. Intl. OFDM Workshop (In-
7 OWo)03, pages 219–224, 2003.
-3 7
10
0 10 20 30 40 [12] D. Petrovic, W. Rave, and G. Fettweis. Phase Noise
E b/N0
Suppression in OFDM using a Kalman Filter. In
Proc. WPMC, 2003.
Figure 9: Performance of the iterative ICI suppression
algorithm for ETSI A channel and δP N = 2 · 10−3 . [13] D. Petrovic, W. Rave, and G. Fettweis. Common
Phase Error due to Phase Noise in OFDM - Estima-
tion and Suppression. In In Proc. of the 15th IEEE
phase noise approximation order. International Symposium on Personal, Indoor and
Acknowledgment Mobile Radio Communications (PIMRC), Septem-
ber 2004.
This work was supported by the German ministry of
research and education within the project Wireless Gi- [14] D. Petrovic, W. Rave, and G. Fettweis. Intercarrier
gabit With Advanced Multimedia Support (WIGWAM) Interference due to Phase Noise in OFDM - Esti-
under grant 01 BU 370 mation and Suppression. In Proc. IEEE VTC Fall,
in press., September 2004.
REFERENCES
[15] L. Piazzo and P. Mandarini. Analysis of Phase
[1] A.G. Armada. Understanding the Effects of Phase Noise Effects in OFDM Modems. IEEE Trans.
Noise in Orthogonal Frequency Division Multi- Commun., 50(10), October 2002.
plexing (OFDM). IEEE Trans. on Broadcasting,
47(2), June 2001. [16] W. Rave, D. Petrovic, and G. Fettweis. Iterative
Correction of Phase Noise in Multicarrier Modula-
[2] G. Caire, G. Taricco, and E. Biglieri. Bit- tion. In In Proc. of the 9th International OFDM
Interleaved Coded Modulation. IEEE Trans. Com- Workshop (InOWo)., September 2004.
mun., 44(3):927–946, May 1998.
[17] P. Robertson and S. Kaiser. Analysis of the effects
[3] R. A. Casas, S.L. Biracree, and A.E. Youtz. Time of phase noise in OFDM systems. In Proc. ICC,
Domain Phase Noise Correction for OFDM Sig- 1995.
nals. IEEE Trans. on Broadcasting, 48(3), Septem-
ber 2002. [18] S.Wu and Y.Bar-ness. Performance Analysis of the
Effect of Phase Noise in OFDM Systems. In IEEE
[4] E. Costa and S. Pupolin. M-QAM-OFDM System 7th ISSSTA, 2002.
Performance in the Presence of a Nonlinear Ampli-
fier and Phase Noise. IEEE Trans. Commun., 50(3), [19] S. Wu and Y. Bar-Ness. A Phase Noise Suppres-
March 2002. sion Algorithm for OFDM-Based WLANs. IEEE
Communications Letters, 44(3), May 1998.
[5] A. Demir, A. Mehrotra, and J. Roychowdhury.
Phase Noise in Oscillators: A Unifying Theory [20] S. Wu and Y. Bar-Ness. A New Phase Noise Mit-
and Numerical Methods for Characterisation. IEEE igation Method in OFDM Systems with Simulta-
Trans. Circuits Syst. I, 47(5), May 2000. neous CPE and ICI Correction. In Proc. MCSS,
Germany, Sep. 2003.
[6] D. J. Higham. An Algorithmic Introduction to Nu-
merical Simulation of Stochastic Differential Equa-
tions. SIAM Review, 43(3):525–546, 2001.
[7] IEEE. Part11: Wireless LAN Medium Access
Control (MAC) and Physical Layer (PHY) Speci-
fications. High-speed Physical Layer in the 5GHz
Band. IEEE Std 802.11a-1999, 1999.
[8] S. Mirabbasi and K. Martin. Classical and modern
receiver arcitectures. IEEE Communications Mag-
azine, pages 132–139, November 2000.