Decision Feedback IBI Mitigation in OFDM
Systems
Wen-Rong Wu
Department of Communication Engineering,
National Chiao-Tung University,
Hsin Chu 300, Taiwan, R.O.C.
E-mail:wrwu@faculty.nctu.edu.tw
Chao-Yuan Hsu
Department of Communication Engineering,
National Chiao-Tung University,
Hsin Chu 300, Taiwan, R.O.C.
E-mail:cyhsu.cm92g@.nctu.edu.tw
Abstract— When the channel response is longer than the guard
interval in OFDM systems, inter-block interference (IBI) will
occur and the desired circular convolution property (the transmit
symbol and the channel response) will not hold. These problems
can seriously degrade the performance of OFDM systems. In
this paper, we combine the decision feedback and cyclic recon-
struction techniques with an iterated channel estimate method
to remedy these problems. We consider pilot-tone based OFDM
systems operating in fading channel environments. We derived a
least-squares (LS) channel estimation and the optimal frequency
domain equalizer (FEQ). Since the computational complexity of
the optimal FEQ is high, we then use a simple suboptimal one.
Incorporating with decision feedback and cyclic reconstruction,
we obtain an iterated IBI mitigation method. Simulations based
on a modified DVB-T specification demonstrate that the proposed
method can effectively reduce the IBI and enhance the overall
system performance.
I. INTRODUCTION
The demand for wireless multimedia services h a s driven the
fast growth of data transmission rate. The high speed transmis-
sion will make the signal more susceptible to channel impair-
ments. A suitable technique to mitigate channel impairments
is the orthogonal frequency division multiplexing (OFDM)
technique. It is a high speed transmission and fading channel
resistance technique, and have been found great potential in
wireline and wireless applications. This includes ADSL, high
data rate wireless LAN applications such as HI PERLAN/2 and
IEEE802.11a/g, and broadcasting services such as DAB and
DVB [1].
The OFDM technique uses the cyclic prefix (CP) to avo id
inter-block interference (IBI) and make the received symbol
be a circular convolution of the transmit symbol and the
channel response. With these properties, we can equalize each
subcarrier signal in OFDM systems by a single complex
weight. If the channel length is greater than the length of CP,
these properties are destroyed and the system performance is
degraded. Some techniques have been proposed to mitigate
this phenomenon. In [2], a decision feedback orthogonality
restoration filter (DFORF) to mitigate the IBI problem in
the time domain was proposed. In this method, a decision
feedback filter is u sed to reduce the IBI and a feedforward
This work was supported by MediaTek research center at National Chiao
T ung University, Hsinchu, Taiwan.
filter is used to restore orthogonality between subcarriers in
the time domain. Since the feedforward filter is obtained
based on MMSE criterion, it requires the noise statistics and
matrix inverse ope rations which will significantly increases the
computational complexity. In [3], an iterated decision feedback
and cyclic reconstruction method was proposed to solve the
problem. However, the channel estimate was conducted using
a blind way which is not applicable in fading channel envi-
ronments.
In this paper, we extend the work in [3] and combine the
method in [3] with a pilot-aided iterated channel estimation
scheme. This scheme is conducted in the time domain using
a least-squares (LS) method. The pilot-aided iterated channel
estimation can provide reliable channel estimates and is more
suitable for wireless fading channels. We also get deeper
inside the problem and derive the optimal frequency domain
equalizer (FEQ). We show that the performance of the iterated
scheme can approach that of an optimal one.
The rest of this paper is organized as follows. Section
II describes the OFDM system model we consider. Section
III presents the proposed iterated signal recovery scheme.
Simulation results are demonstrated in Section IV. Finally, we
draw some conclusions in Section V.
II. S
IGNAL MODEL
Fig. 1 depicts the baseband block diagram of an OFDM
system. In such a system, the available bandwidth B is di-
vided into
equally spaced subcarriers. Each subcarrier has
bandwidth
,where
is the symbol
sampling period. The operation of the system can be described
as follows. First, source bits are mapped into a symbol
modulating the
subcarrier, where
is the OFDM symbol
(block) index, and
is the subcarrier index. The series data
symbols are passed to a series-to-parallel converter (S/P) and
inverse fast Fourier transform (IFFT) to obtain the modulated
signal. Then, a CP with length
is inserted b etween the
consecutive OFDM symbols. The CP is made up of the last
samples of the
-point IFFT outputs. Finally, the data
are passed through a parallel-to-series converter (P/S) and
transmitted. We assume that the channel is a multipath channel
and noise is additive white Gaussian noise (AWGN). Also, we
assume that the channel delay spread is
which is longer than
26190-7803-8834-8/05/$20.00 ©2005 IEEE.