A New Inter-Carrier Interference Cancellation

Using CP-ICA Scheme in OFDM Systems

Jong-Deuk, Kim

Youn-Shik, Byun

Dept. of Electronics Engineering, University of Incheon

177 Dohwa-Dong, Nam-Gu,
InCheon 402-749, Korea
jdkim@incheon.ac.kr

Dept. of Electronics Engineering, University of Incheon

177 Dohwa-Dong, Nam-Gu,
InCheon 402-749, Korea
byun@incheon.ac.kr

Abstract A well-known problem of OFDM is its sensitivity to

frequency offset which leads to Inter-Carrier Interference (ICI)
in the OFDM symbol. This ICI causes severe degradation of the
BER performance of the OFDM receiver. In this paper, we
propose a new ICI cancellation algorithm which estimates
frequency offset at the time-domain by using the Cyclic Prefix
(CP) based Independent Component Analysis (ICA) method to
the received subcarriers. We derive the frequency offset
estimation technique by using CP-ICA analyzing the phase
rotation of the OFDM symbols. This algorithm is based on a
statistical blind estimation method, which mainly utilizes the
Eigenvalue Decomposition (EVD), rotating phasor and the 4th
cumulants. In the case of 0.0< <1.0, simulation results show
that the proposed frequency offset estimator is more accurate
than the other estimators under AWGN and Rayleigh flat fading
channels.
Keywords-OFDM; Frequency Offset; ICI; CP-ICA;

I.

INTRODUCTION

Orthogonal Frequency Division Multiplexing (OFDM) is

emerging as the preferred modulation scheme in modern high
data rate wireless communication systems. A well known
problem of OFDM, however, is its sensitivity to frequency
offset between the transmitted and received signal, which may
be caused by Doppler shift in the channel, or by the difference
between the transmitter and the receiver local oscillator
frequencies. The carrier frequency offset causes loss of
orthogonality and then the signals transmitted on each carrier
are not independent on each other, thus leading to the ICI [1].
Therefore, a synchronization problem of carrier frequency
offset is very crucial for possible application of OFDM in high
data rate transmission wireless communications. The existing
approaches that have been developed to reduce ICI can be
categorized as the one using pilot symbols, the blind scheme
using CP, and ICI self-cancellation (SC) scheme [2]. Most
existing frequency offset estimation techniques in OFDM
systems require the pilot symbols [3]-[5]. However, the use of
pilot symbols reduces the bandwidth efficiency because the
pilots occupy some part of the useful bandwidth. Therefore, a
blind approach scheme for frequency offset estimation was
introduced [6]. However, the algorithm relies on virtual
subcarriers, which also lower the bandwidth efficiency. For

1550-2252/$25.00 2007 IEEE

more realistic synchronization, some methods have been

proposed for blind estimation of timing offset as well as
frequency offset [7]-[9], [13]. These algorithms mainly utilize
the CP and some methods [8] require the OFDM signal to be
cyclostationary. In addition, statistical separation approaches
have also been explored to do a blind source separation [10][13].
In this paper, we propose a blind method to estimate
frequency offset in OFDM systems. The proposed method
uses CP-ICA approach that statistically estimates the
frequency offset and corrects the offset using the estimated
value at the receiver. The paper is organized as follows:
Section describes the standard system model and the
problem of ICI in OFDM. Section describes the
Independent Component Analysis(ICA) scheme and the
proposed CP-ICA methods in this paper. Section provides
a simulation result that tests the BER performance and also
compares it with the SC, ML and CP-ML[2, 4, 7]. Finally,
concluding remarks are presented in section .
II.

SYSTEM MODEL AND INTER-CARRIER INTERFERENCE

PROBLEM IN OFDM SYSTEMS

A typical discrete-time baseband OFDM transceiver system

is shown in Fig. 1 [14].

Figure 1. Baseband OFDM system

The N symbols are mapped to bins of an inverse fast

Fourier transform (IFFT).
xl (n ) =

1
N

N 1

m =0

X l (m )e

2 nm
N

(1)

where Xl (m) is the baseband l-symbols on m-th subcarrier.

The demodulated symbol stream is given by

2369

Yl ( k ) =

N 1

yl (n )e

2 kn
N

(2)

+ W l (k )

use the eigenvalue decomposition of covariance matrix.

n=0

[ E v , D v ] = e ig ( { Y Y T } ) = E v D v E v T

where Wl (k) corresponds to the FFT of the samples of wl

(n), which is the Additive White Gaussian Noise(AWGN)
introduced in the channel.
In this paper, the frequency offset is modeled as a
multiplicative factor introduced in AWGN channel, as shown
in Fig. 2.

where Ev is the orthogonal matrix of eigenvectors of

eig ( E{YY T }) and Dv is the diagonal matrix of its eigenvalues.
The whiteness or uncorrelatedness is a necessary condition for
the stronger statistically independent condition [10 - 11].

O = Ev Dv (1/ 2) Ev T , Y = OY = OAX = RX

Figure 2. Frequency Offset Model under AWGN Channel

The received OFDM signal is given by

yl (n ) = xl (n )e

2 n
N

(3)

+ wl (n)

where (= foffset / f) is the normalized frequency offset,

foffset is the frequency difference between the transmitted and
receiver frequencies and f is the subcarrier spacing. wl (n) is
the AWGN introduced in the channel. The effect of this
frequency offset on the received symbol stream can be under stood by considering the received symbol Yl(k) on the k-th
subcarrier.
Yl ( k ) =

N 1

y ( n)e
l

i 2 kn / N

, k = 0,1, ...., N 1

1
=
N

N 1

m=0

N 1

( m ) e

= X l ( k ) I l (0) +

2 n (m+ k )
i
N

m = 0 ,m k

= 1 y (n) : cyclic prefix parts

l
1 n
: useful data parts
=L
[2(N /4)] =
,
N 1
= y (n) L : CP length(= N / 4)
l
2
th
n=NL
yl (n) : received l symbols
1

X l ( m )I l ( m k ) + W l ( k )

2
I

T
[ E , D ] = eig (
len gth (

(10)

Step 2. Centering and Whitening

r

(O r )
is centering matrix,

= { } , =

(11)

where r
is statistically
independent matrix and {}
denotes the expectation operator.

To rigorously define ICA, we can use a statistical model

[10-12]. We have an observable random vector y (n ) .
y ( n) = aij x jn ( n), n = 0,1,...., N 1

T
1
2
), P = ED E , O = P
)

where E is eigenvectors, D is eigenvalues, P is whitener

matrix, and O is pseudo-inverse of P

(5)

(9)

Step 1. Polar Decomposition

FREQUENCY OFFSET ESTIMATION USING CP-ICA

I 1 I 1

(8)

Therefore, the unmixing matrix can be constructed as =

UO = A-1. The proposed ICI cancellation using CP-ICA must
be needed to estimate Carrier Frequency Offset (CFO) in
OFDM systems. The CFO estimation performs as follows:
i (i=1,2) are linear mixtures of statistically independent each
other by the frequency offset and AWGN.

+ Wl (k )

The first term in (4) denotes the useful component. The

transmitted symbol Xl(k) is rotated and attenuated. The second
term is the complex coefficients for the ICI components in the
received signal. For sufficiently large N, ICI can be modeled as
a Gaussian random process with zero mean variance by the
central limit theorem. Therefore, we can estimate the frequency
offsets from the estimate of phase rotation.
III.

U Y = U ( R X ) = U ( O A X ) = U O ( A X ) = X

n =0

N 1

(7)

where O is whitening matrix, new vector Y is white. i.e., its

components are uncorrelated and their variance equal unity.
The whitening transforms the mixing matrix A into a new
rotating matrix R ( = O A ) .
We just need to find an orthogonal transformation U to
make components of X = U Y .

(4)

n =0

(6)

Step 3. The 4th Cumulants of ICA [12]

i = 0 j =0

yy

= { ( T )} , y y = { ( H )}

C = yy yy [ 0 , 1 ] yy [ 2 , 3 ]
yy [ 0 , 2 ] yy [ 1 , 3 ]

y(n) is assumed to be linear mixtures of some mutually

statistically independent variables aij. Mathematically, Y is
generated according to Y=AX. The matrix A is mixing matrix.
In order to estimate unmixing matrix (= A-1), whitening
processing is needed. One popular method for whitening is to

(12)
(13)

yy [ 0 , 3 ]conj ( yy [ 1 , 2 ])

where T is transpose, H is Hermitian, C is the 4th

cumulants and i is delay factor.

2370

IV.

Step 4. Extraction of important eigenvector from Step 3

(14)

[ E 1 , D 1 ] = e ig ( C 2 )
i = a rg m a x ( d ia g { D 1 } ) , A = {1, 2 , ... , n }
i A

j = a rg m a x ( d ia g { D 1 } )
j A , j i

where i is 1st and j is 2nd index of max eigenvalues.

Step 5. Buffering with respect to important eigenvector
m1 = m1[ MK ] = E1 (i)
, B = [ m1 m2 ], M = K = 2

m2 = m2[M K ] = E1 ( j)

(15)

where B is buffer with respect to max eigenvectors matrix.

Step 6. Rotating to Step 5
max( E1(i)) min( E1 (i)) max( E1 ( j )) min( E1 ( j ))

g=
m1 (1, K )
m2 (1, K )

m1( M ,1)
m2 ( M ,1)

(16)

SIMULATION RESULTS

In order to compare the three different ICI cancellation

schemes, we will assume that timing synchronization is
perfect for all methods at the receiver. The OFDM transceiver
system was implemented as specified by Fig. 1. Frequency
offset was introduced as the phase rotation as given in (3). In
the first simulation, proposed frequency offset estimator was
implemented as specified by Fig. 2. The channel is assumed
AWGN. The OFDM parameters chosen are as follows;
modulation schemes: uncoded 16 / 64-QAM, the number of
subcarriers: N = 256, CP length: L = N/4, and normalized
frequency offsets: = {0.037, 0.47, 0.67, 0.97}. The
simulation is carried out over 1000 OFDM symbols for each
SNR values. The BER curve for standard OFDM system
without ICI cancellation is shown in Fig. 3.

cos
R=
sin

sin c s
=
cos s c
1 0
T
T
[ E2 , D2 ] = eig
, G = E2 ( g g ) E2
0 R

where g is real and imaginary axis separation to the max

eigenvectors(=B), R is basis rotating matrix, G is rotated
matrix of independent components.
Step 7. Extracting max eigenvector from the G matrix and
obtaining basis rotating matrix (=R).
(17)
[ E3 , D3 ] = eig (G )

Figure 3. BER performance without ICI cancellation of a standard OFDM

k = arg max(diag{D3}), A = {1, 2,..., n}

kA

(2) 1 (3)
1 (1)
= E3 (k ), c =
+
, s=
2
2
2c

where k is the index of max eigenvalues and is max

eigenvectors.
Step 8. Update of basis rotating matrix.
c conj ( s )
1 0
Ru =
, U b = 0 1 , U = U b Ru
s
c

(18)

where Ru is updated basis rotating matrix, Ub is basis

Unitary, U is update rotated unitary matrix.
Step 9. Comparison for the optimal rotated matrix
If | s | > Th [ go Step 5. updating buffer]

(19)
Else if | s | < Th [ obtain the optimal rotated matrix]
(20)
[ M K ] = (U O )
B = R

End
where Th is decision threshold (=1 / No. of subcarriers) and
is the optimal rotated square matrix.

Step 10. Frequency Offset Estimation

M :(M)th rowvector

 = tan1 ( M conj( M1)) /2 ,

th
M 1 : (M 1) rowvector

if ( <106 ),  =1.0 + , end

(21)

In the case of larger alphabet sizes and larger frequency

offsets, OFDM systems do deteriorate the performance too
greatly. Therefore, we can conclude that larger alphabet sizes
and larger frequency offsets are more sensitive to ICI.
Fig. 4 and Fig. 5 provide comparisons of the performance of
the SC[2], ML[4], CP-ML[7] and CP-ICA schemes for
different alphabet sizes and different values of the frequency
offset in 0.0< <0.5. In the presence of very small frequency
offset and small alphabet size, the SC gives the best results.
However, for larger alphabet size and larger frequency offset
such as 64-QAM and frequency offset of 0.47, the SC and ML
method does offer a serious decrease in performance. But the
CP-ML and CP-ICA method gives the best in the overall
results.
Fig. 6 and Fig. 7 show comparisons of the performance of
the SC, ML, CP-ML and CP-ICA schemes for different
alphabet sizes and different frequency offsets in 0.5< <1.0.
We can observe the BER error floor. The SC, ML and CP-ML
methods are very hardly depredated BER. Since these
techniques do not completely cancel the ICI from adjacent
subcarriers and the effect of this residual ICI increases for
larger alphabet sizes and frequency offset values. However, by
using the proposed CP-ICA method for high normalized
frequency offset (= ), we can compensate the frequency drift
caused by the difference between oscillators in the transmitter
and the receiver. This is attributed to the fact that the CP-ICA
method estimates the frequency offset very accurately and

2371

cancels the frequency offset. So, the proposed CP-ICA method

has an extreme performance overall in 0.0< <1.0. It gives a
significant boost to performance under AWGN channel.

It is assumed that path gains are constant during a frame

(quasi-static flat fading) and vary from one frame to another
frame. Also, we will assume that channel impulse response is
known perfect for all methods at the receiver.

Figure 8. Frequency Offset Model under Rayleigh Flat fading Channel

The received OFDM signal is given by

Figure 4. BER Performance with ICI Cancellation, = 0.037

y l ( n ) = hl ( n ) * x l ( n ) e

2 n
N

(22)

+ wl ( n )

where * denotes convolution. After CFO estimation using

CP-ICA, we compensate CFO in the time-domain, do FFT and
generate demodulated symbol stream which are shown below.
N 1

yl (n) = yl (n)ei 2 n / N , Yl (k) = ( yl (n))ei 2 kn/ N

n=0

(23)

X l (k) = Yl (k )/ H l (k)

Figure 5. BER Performance with ICI Cancellation,

= 0.47

Fig. 9 BER Performance with ICI cancellation,

Figure 6. BER Performance with ICI Cancellation,

= 0.037

= 0.67

Fig. 10 BER Performance with ICI cancellation,

Figure 7. BER Performance with ICI Cancellation,

= 0.97

In the second simulation, the proposed frequency offset

estimator is implemented as specified by Fig. 8 in multi-path
fading channels. The channel is assumed Rayleigh flat fading.

= 0.47

Fig. 9 and Fig. 10 provide comparisons of the performance

of the SC, ML, CP-ML and CP-ICA schemes for uncoded
16/64-QAM and different values of the frequency offset in
0.0< <0.5.
In the presence of very small frequency offset ( = 0.037)
under Rayleigh flat fading channels, the CP-ML and CP-ICA

2372

method gives the best results. However, the SC and ML

methods does offer error floor in performance. In the case of
larger frequency offset ( = 0.47) under Rayleigh flat fading
channels, Fig. 10 shows that the SC and ML method does
offer error floor in performance. However, the CP-ICA and
CP-ML method gives the better results than the other methods.

0.0< <0.5. However, this method performs at frequency

domain for estimation of the frequency offset.
In this paper, we propose a new ICI cancellation using CPICA scheme in OFDM systems. This method is a blind
scheme. Since our scheme does not need any pilot symbol in
estimation, we can expect enhanced bandwidth efficiency in
OFDM systems. We derive the frequency offset estimation
technique by using CP-ICA analyzing the phase rotation of the
OFDM symbols. On the whole, the proposed new CP-ICA
scheme is very accurately estimated even at larger alphabet
sizes and larger frequency offsets. In the case of 0.0< <1.0,
simulation results show that the proposed frequency offset
estimator is more accurate than the other estimators under
AWGN and Rayleigh flat fading channels.
REFERENCES

Fig. 11 BER Performance with ICI cancellation,

= 0.67

[1]

[2]

[3]

[4]

[5]

Fig. 12 BER Performance with ICI cancellation,

= 0.97

Fig. 11 and Fig. 12 show comparisons of the performance

of the SC, ML, CP-ML and CP-ICA schemes for different
frequency offsets (0.5< <1.0) under Rayleigh flat fading
channel. The SC, ML and CP-ML methods are given BER
error floor. Because the scheme compensates the frequency
drift exactly, the proposed CP-ICA method gives the best
extreme performance than the other schemes.
Therefore, we can see that the proposed CP-ICA method
outperforms the other methods in 0.0< <1.0 under Rayleigh
flat fading channels.
V.

CONCLUSION

The ICI is a serious problem in OFDM system especially in

existing frequency offset and mobile scenarios. The SC [2]
does not require very complex hardware or software for
implementation. However, it is not bandwidth efficient as
there is a redundancy of twice for each carrier. The ML [4]
method also introduces the same level of redundancy but
provides better BER performance, since it accurately estimates
the very small frequency offset. Its implementation is more
complex than the SC method. On the other hand, the CP-ML
[7] method does not reduce bandwidth efficiency as the
frequency offset can be estimated from the CP in each OFDM
frame. Also, it performed the best of the three methods in

[6]

[7]

[8]

[9]

[10]
[11]

[12]
[13]

[14]

2373

T. Pollet, M. van Bladel and M. Moeneclaey, BER sensitivity of

OFDM systems to carrier frequency offset and Wiener phase noise,
IEEE Trans. on Communications, vol. 43, no. 2, pp. 887-895, Feb. 1995
Y. Zhao and S. Hggman, Intercarrier interference self-cancellation
scheme for OFDM mobile communication systems, IEEE Trans. on
Communications, vol. 49, no. 7, pp. 1185 1191, July 2001.
A. J. Coulson, Maximum likelihood synchronization for OFDM using a
pilot symbol: algorithms, IEEE J. Sel. Areas Communications, vol. 19,
no. 12, pp. 24862494, Dec. 2001.
P. H. Moose, A Technique for Orthogonal Frequency Division
Multiplexing Frequency Offset Correction, IEEE Trans. on
Communications, vol. 42, no. 10, pp. 2908-2914, Oct. 1994.
H. Chen and G. J. Pottie, A comparison of frequency offset tracking
algorithms for OFDM, Proc. of the IEEE Global Communication
Conference, vol. 2, pp. 1069-1073, December 2003.
U. Tureli and H. Liu, Blind carrier synchronization and channel
identification for OFDM communications, Proc. of IEEE Int. Conf. on
Acoustics, Speech and Signal Processing, pp. 3509-3512, Seattle, WA,
May 1998.
J. Beek, M. Sandell, P. 0. Borjesson, ML estimation of time and
frequency offset in OFDM systems,
IEEE Trans. on Signal
Processing, vol. 45, no. 7, pp. 1800-1805, July 1997.
H. Bolcskei, Blind estimation of symbol timing and carrier frequency
offset in pulse shaping OFDM systems, Proc. of IEEE Int. Conf. on
Acoustics, Speech and Signal Processing, pp. 2749-2752, Phoenix, AZ,
Mar. 1999.
J. H. Gunther, H. Liu and A. L. Swindlehurst, A new approach for
symbol frame synchronization and carrier frequency estimation in
OFDM communication, Proc. of The IEEE Int. Conf. on Acoustics,
Speech and Signal Processing, pp.2725-2728, Phoenix, AZ, March 1999.
J. F. Cardoso, "Source Separation using higher order moments", In Proc.
ICASSP'89, pp. 2109-2112, May 1989.
P. Comon, P. Chevalier and V. Capdeville, Performance of contrastbased blind source separation, In proc. SPAWC'97, vol. 1, no. 1, pp.
345-348, April 1997.
A. Hyvarinen, J. karhune, E. Oja, Independent Component Analysis,
Wiley Interscience, 2001. pp. 40-42.
C. S. Wong and D. Obradovic, "Independent Component Analysis for
blind equalization of frequency selective channel," IEEE XIII Workshop
on Neural Networks for Signal Processing, pp. 419- 427, Sep. 2003.
R. Prasad, OFDM for Wireless Multimedia Communications, Artech
House, 2004. pp. 83-116.