You are on page 1of 5

A Conjugate Operation for Mitigating Intercarrier Interference of OFDM Systems

Yuan-Kwei Chang
EECS Department
University of California, Irvine
Irvine, CA 92697
E-mail: ykchang1107@yahoo.com

Hen-Geul Yeh
Electrical Engineering Department
California State University, Long Beach
Long Beach, CA 90840-8303
E-mail: heyeh@csulb.edu
Abstract In order to mitigate the intercarrier
interference (ICI), a two-path algorithm is developed
for orthogonal frequency division multiplexing
(OFDM) systems. The first path is a regular OFDM
system for transmission and receiving. The second path
requires a conjugate operation of the 1st block of data to
form the 2nd block of data for transmission at the
transmitter. The received 2nd block data will be
conjugated at the receiver and then demodulated. The
demodulated results of the 1st path are combined with
that of the 2nd path with equal weights to form the final
detected data symbols. This conjugate cancellation
(CC) scheme provides (1) a high signal to ICI power
ratio in the presence of small frequency offsets (33 dB
higher than that of linear self-cancellation [1-2]
algorithms, at fT = 0.1% of subcarrier frequency
spacing); (2) backward compatibility with the existing
OFDM system; (3) significantly better BER
performance in both additive white Gaussian noise
(AWGN) and frequency selective fading channels.
I.

INTRODUCTION

Lately, researchers in universities and manufacturers


in industry show great interest in orthogonal
frequency division multiplexing (OFDM) systems.
OFDM has already been accepted for the new wireless
local area network standards as IEEE 802.11. OFDM is
very robust to wireless channel impairments (e.g. multipath propagation and frequency-selective fading). The
receiver design is simplified due to less complex
equalization. The OFDM system has high spectral
efficiency due to overlapping subcarrier spectra.
However, one of the major disadvantages of such a
multi-carrier modulated system is the sensitivity of its
performance to synchronization error, such as frequency
or phase offsets. The frequency offset can result from a
Doppler shift due to a mobile environment, as well as
from a carrier frequency synchronization error. Such
frequency offsets cause a loss of the carriers

0-7803-8521-7/04/$20.00 2004 IEEE

orthogonality, and hence create intercarrier interference


(ICI). Currently, three different approaches for
mitigating ICI have been proposed including: ICI selfcancellation [1-2], frequency-domain equalization [3],
and time-domain windowing scheme [4]. Frequency
offset estimation techniques using training sequence
such as pilot symbols are proposed in [5-6]. This study
focuses on the ICI cancellation scheme and its effect on
OFDM communication systems in AWGN and fading
channels, assuming that the synchronization, including
phase, frequency, and timing has done by using
repeated preamble sequence, but the ICI may still exist
due to the frequency offset estimation error or
unexpected Doppler velocity.
This paper is organized as follows. The OFDM
system in additive white Gaussian noise (AWGN)
channel is discussed in Section II. The math model of
the transceiver is described in Section III, along with a
discussion of the weighting function of the data symbol
due to frequency offset. Section IV presents a CC
scheme and the corresponding architecture of the
transceiver. Simulation results are discussed in Section
V. Conclusions are given in Section VI.
II.
OFDM SYSTEMS
A. Regular OFDM Transmitter
A conventional OFDM modulation is employed at the
transmitter. The baseband transmitted signal x k at the
output of the IFFT can be written as
N 1

x k = d n e

2
nk
N

k = 0,1,2,..., N 1

(1)

n =0

where

2
nk
N

dn

is

the

data

symbol,

and

, k = 0,1, ..., N 1 , represent the corresponding

orthogonal frequencies of N subcarriers. Clearly, a


group of N different data symbols is mapped onto N
subcarriers via the IFFT processor. Note that the IFFT

3965

0-7803-8521-7/04/$20.00 (C) 2004 IEEE

will have TOFDM seconds to complete its operation. The


duration TOFDM for an OFDM symbol is NTs, where Ts
is the data symbol time duration. For simplicity, T is
used to replace TOFDM in the following discussion.
B. Regular Receiver Baseband Processing
At the receiver, the OFDM signal is mixed with a local
oscillator signal. Assuming it is f above the carrier
frequency of the received OFDM signal due to
frequency estimation error or Doppler velocity, the
baseband FFT demodulator output is then given by

1
d n =
N

N 1

r e

2
nk
N

n = 0,1,2,..., N 1

(2)

k =0

where rk = x k e

2
kfT
N

+ wk represents the received

signal at the input to the FFT processor, wk

is the

where

1 1 e j 2 ( m n + fT )
(5)
2
j ( m n + fT )
N
1 e N
Clearly,
the
complex
weighting
functions
u 0 , u1 , ..., u N 1 indicate the contribution of each of the
u m n =

N data symbols d m to the FFT output d n . Those terms


that mn, they represent the cross talk from the
undesired data symbols. The weighting function of the
IFFT (d n ) FFT (rk ) pair, the transmitter-receiver
operation, is a periodic function with a period equal to
N. If the normalized frequency offset fT equals zero,
With some
then d n is exactly equal to d n .
manipulation, equation (5) can be rewritten as

AWGN, and d n is the output of the FFT processor.


j

u m n =e

( N 1)( m n + fT )

kfT

The
e N
, k = 0,1, ..., N 1 , represents the
corresponding frequency offset of the received signal at
the sampling instants, and fT is the frequency offset
to subcarrier frequency spacing ratio.
III.
ANALYSIS
The ICI resulting from the carrier frequency offset is
analyzed by using both mathematically in discrete
domain and graphically with continuous curves in this
section.
Following the similar approach in [2],
expressions are derived for each demodulated
subcarrier at the receiver in terms of each transmitted
subcarrier and N complex weighting functions.
Without loss of generality, the noise wk in the received
signal is ignored in the following discussion.
Substituting (1) into (2) and after some manipulation, it
can be shown that
N 1

N 1

j
1
dn = d m e
N m=0 k =0

2k
( m n + fT )
N

n = 0,1,2,..., N 1

(3)

Equation

(6)

( N 1)( m n + fT )

consists
and

N 1

j 2 ( m n + fT )

N 1

= d m um n

rotation

Dirichlet

factor
function

at N = 8 , n = 0 , and fT = 0 , the range of m is set


from zero to eight in Fig. 1. Note that there are small
approximately anti-symmetrical regions around the
zero-crossing points at the integer indexes m = 1, 2, 6
and 7 for the real weighting function. Similarly, there
are small approximately anti-symmetrical regions
around the zero-crossing points at the integer indexes m
= 2, 3, 4, 5 and 6 for the imaginary weighting function.
These small approximately anti-symmetrical regions
around the zero-crossing points at the majority integer
indexes are very useful for ICI cancellation as
explained in the following sections.

(4)

n = 0,1,2,..., N 1

m =0

0-7803-8521-7/04/$20.00 2004 IEEE

the

of

(6)

1 sin( (m n + fT )) . To illustrate a complete cycle

N
sin( (m n + fT ))
N

Taking the advantage of the properties of geometric


series, it can be derived as

1
1 e
dn = d m
2
j
( m n + fT )
N m =0
1 e N

1 sin( (m n + fT ))

N
sin( (m n + fT ))
N

3966

0-7803-8521-7/04/$20.00 (C) 2004 IEEE

multiplexing (TDM), or code division multiplexing


(CDM), or dual antennas. At the receiver, the conjugate
algorithm requires a conjugate operation on the received
signal first, and then performs the FFT operation as
defined in (8):

1
Real

Weighting function of regular OFDM

0.8
Imag
0.6
0.4

1
d n' =
N

0.2
0

N 1

2
j
nk
' *
N
k

(r ) e

signal, wk'
output of

-0.4
-0.6
-0.8
0

3
4
Normalized index m

Fig. 1. The weighting function.


III. CONJUGATE ALGORITHM
To mitigate the impact of ICI, a new conjugate
algorithm is developed. The basic idea is to add a
simple circuitry that provides weighting factors with
opposite polarities. To do this, this new algorithm
should produce the same weighting curves as that of
Fig. 1. But these curves shift to the right (instead of
left) by fT when the frequency offset fT is greater
than zero. This shift-to-right operation changes the
weighting functions from zero to negative or positive
values in the reversed direction of that of the shift-toleft operation. The combined weighting functions
(regular and conjugate algorithms) will provide a
significantly smaller weighting function on undesired
data symbols while maintaining the same weighting
function on the desired data symbol for small frequency
offsets as illustrated in the next three subsections.

2
j
kfT
N

(8)

k =0

where rk' = x k' e

-0.2

n = 0,1,2,..., N 1

2
kfT
N

+ wk' represents the received


is the independent AWGN, and d n' is the
the

FFT

processor.

The

term

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

represents
the
corresponding frequency offset of the received signal at
the sampling instants. Without loss of generality, the
noise wk' is assumed to be zero in the following
discussion. Substituting (7) into (8) and after some
manipulation, it can be shown that

1
d n' =
N

N 1

N 1

m =0

k =0

dm e

2k
( m n fT )
N

n = 0,1,2,..., N 1

(9).

By following a similar derivation as that of Section III,


one obtains the weighting functions for the FFT output:

(10)
1 sin( (m n fT ))

N
sin( (m n fT ))
N
Equation (10) is similar to (6), but the sign of the
frequency offset term, fT , is changed from positive
to negative.
This weighting function of the
*
(IFFT (d n )) FFT (rk' )* pair, conjugate transmitter
receiver operation, is identical to the weighting function
A. The Conjugate Algorithm
At the transmitter, this algorithm requires a conjugate of (6) at fT = 0 . On the other hand, the frequency
offset fT > 0 , will result a shift to the right operation
operation on the IFFT output as defined in (7):
*
2
(7) on the weighting function of (10) as opposed to a shift
j nk
N 1 j 2N nk N 1
*
'
N
to the left of (6).
= (d ) e
x = d e
k = 0,.., N 1
k

n=0

v m n =e

( N 1)( m n fT )

n=0

B. Conjugate Cancellation Scheme


and Assuming that both outputs of a regular
2
j
nk
IFFT (d n ) FFT (rk ) pair OFDM system and a
e N , k = 0,1, ..., N 1 , represent the corresponding
*
' *
pair OFDM
orthogonal frequencies of N subcarriers. Note that in conjugate (IFFT (d n )) FFT (rk )
order to demodulate the original signal xk and the system can be combined without interfering each other
conjugate signal xk separately, xk needs to be transmitted at the receiver by using a division multiplexing
independently. This can be achieved by using frequency technique, the final detected symbol is then chosen as
division multiplexing (FDM), or time division
where

dn

is

the

data

symbol,

0-7803-8521-7/04/$20.00 2004 IEEE

3967

0-7803-8521-7/04/$20.00 (C) 2004 IEEE

the averaged detected symbols of the regular OFDM


receiver and the conjugate algorithm as follows:

n = 0,1,2,..., N 1 .

100
SICI ratio (dB), Regular, N=16, Conj, const and linear

1
d = (d n + d n' )
2
"
n

110

(11)

This is the CC scheme. In fact, it is a simple version of


time space coding if dual antennas are employed [7].
The signal to ICI power ratio (SIR) of the CC
algorithm, as a function of frequency offsets, is plotted
in Fig. 2. The SIR of a regular OFDM system is
independent to N. But the SIR of the CC algorithm is a
function of N for large frequency offsets. For a small
frequency offset, the SIR of the CC algorithm is about
the same for different N. It shows that the SIR of the
CC algorithm is about 17 dB higher than that of the
regular algorithm at 4% frequency offset. On the other
hands, the SIR of the CC algorithm is smaller than that
of the regular OFDM algorithm at fT > 0.25 . Fig. 3
depicts the SIR for four different systems: regular
OFDM, self-cancellation schemes with constant and
linear components of ICI [1-2], and this CC scheme at
N = 16. It is shown that this CC algorithm has a highest
SIR than others when frequency offsets are small. A
single antenna architecture for the transmitter and
receiver is described in Figs. 4(a) and (b), respectively.
A CDM, TMD, or FDM is required at the parallel
transmitter to make parallel transmission. A demultiplexing circuit is required at the receiver in order
to perform parallel receiving operations accordingly. A
2N-point FFT is applied for both path data for efficient
computation.

90
CC
80
70
Linear Self-Cancellation
60
50

Constant Self-Cancellation

40
30
20
Regular
10
0

0.02

0.04

0.06

0.08
0.1
0.12
Normalized Frequency index

0.14

0.16

0.18

0.2

Fig. 3. SIR of four algorithms vs. frequency offsets.


f1

Modulation
(BPSK,
dn
QPSK,
S/P
16QAM,
64QAN)

N-Point
IFFT

P/S

1st path
TDM

( )*

2nd path

Fig. 4(a). The parallel transmitter.


f1

N Even Bins

dn

1st path

S/P

Demux

2NPoint
FFT

P/S

Demodulation

( )*

60

2nd path
SICI ratio (dB), Regular, Conj parallel Combined

50

Fig. 4(b). The parallel receiver.

40

IV. SIMULATION RESULTS


30

20

N=8

10

N=16

Regular

N=32

-10
N=128
-20
0

0.05

0.1

0.15

0.2
0.25
0.3
Normalized Frequency index

0.35

0.4

0.45

0.5

Fig. 2. SIR of the CC algorithm vs. frequency offsets.

0-7803-8521-7/04/$20.00 2004 IEEE

The signal processing for the transmitter and receiver


described in Figs. 4(a) and 4(b), respectively, was
simulated in AWGN channel. For fair comparison,
each branch at the transmitter of the CC system is at
half of the original signal power of the regular OFDM
system. At N = 16, the BER performance of 16-QAM
for a regular OFDM and a CC OFDM in AWGN
channel is provided in Fig. 5. At fT = 0.04 , it shows
that the performance of the CC algorithm is better than
that of the regular algorithm by 14 dB at BER =10-6.

3968

0-7803-8521-7/04/$20.00 (C) 2004 IEEE

BER of 16-QAM OFDM in AWGN

1.E+00

V. CONCLUSIONS

Theo-AWGN

By using the CC scheme, the sensitivity of the OFDM


system to ICI is reduced. The key feature of this new
scheme is that it provides a much better SIR over the
regular OFDM system in the presence of small frequency
offset. This CC scheme provides an excellent BER
performance for small frequency offset over a regular
OFDM system in AWGN channel and 3-ray Rummler
fading channels. Although the bandwidth efficiency is
reduced to half due to twice transmission, it can be
compensated by using larger signal alphabet sizes. This
CC scheme can also be considered as a simple space
coding with dual antennas as in [7].

Reg-0.02

1.E-01

CC-0.02

1.E-02

Reg-0.04

1.E-03

Reg-0.06

CC-0.04
CC-0.06

1.E-04
1.E-05
1.E-06
0

12

16

20

24

28

Eb/No dB

REFERENCES

Fig. 5. BER performance in AWGN channel.


A 3-ray Rummler fading model [8] with frequency
offset is employed. A 25% guard time is applied to the
DQPSK OFDM signal with N=1024. Frequencydomain differential coding is applied in both systems in
order to avoid channel response estimation. The data
rate is 20 Mbps. This channel model is composed of
three propagation paths: a line-of-sight (LOS) path, and
two specular reflections (2nd and 3rd rays) with Rayleigh
distribution. The 2nd ray has a signal strength that is
83% of that in the LOS path with a 50 ns delay. The
amplitude of the 3rd ray is 2% of that of the LOS ray
with a 150 ns delay. Fig. 6 shows that the BER
performance of the CC system is significantly better
than that of regular system when the frequency offset
becoming a dominate factor, especially at high SNR.
Reg-0.01

1.E+00

CC-0.01

BER of DQPSK OFDM

Reg-0.04
1.E-01

CC-0.04

1.E-02

CC-0.1

Reg-0.1

1.E-03
1.E-04
1.E-05
4

12

16

20

24

28

32

36

40

Eb/No (dB)

Fig. 6. BER performance in fading channel.

0-7803-8521-7/04/$20.00 2004 IEEE

[1] Y. Zhao and S-G. Haggman, Intercarrier


interference self-cancellation scheme for OFDM
mobile communication systems, IEEE Trans.
Commun., vol. 49, no. 7, pp. 1185-1191, July 2001.
[2] J. Armstrong, Analysis of new and existing methods
of reducing intercarrier interference due to carrier
frequency offset in OFDM, IEEE Trans. Commun.,
vol. 47, no. 3, March 1999, pp. 365-369.
[3] N. A. Dhahi et al., Optimum finite-length
equalization for multicarrier transceivers, IEEE
Trans. Commun., vol. 44, no. 1, pp. 56-64, Jan. 1996.
[4] C. Muschallik, Improving an OFDM reception using
an adaptive Nyquist windowing, IEEE Trans.
Consumer Electron., vol. 42, pp. 259-269, Aug.
1996.
[5] P. H. Moose, A technique for orthogonal frequency
division multiplexing frequency offset correction,
IEEE Trans. Commun., vol. 42, no. 10, Oct. 1994,
pp. 2908-2914.
[6] T. M. Schmidl and D. C. Cox, Robust frequency
and timing synchronization for OFDM, IEEE
Trans. Commun., vol. 45, no. 12, pp1613-1621,
Dec. 1997.
[7] S. M. Alamouti, A simple transmit diversity
techniquefor wireless communications, IEEE J.
Select. Areas Commun., vol. 16, no. 10, pp. 14511458, Oct. 1998.
[8] M. Rice, A. Davis, C. Bettweiser, A wideband
channel model for aeronautical telemetry, IEEE
Trans. Aerospace and Electronics Systems, vol. 40,
pp. 57-69, Jan. 2004.

3969

0-7803-8521-7/04/$20.00 (C) 2004 IEEE

You might also like