256

IEEE TRANSACTIONS ON SIGNAL PROCESSING, VOL. 55, NO. 1, JANUARY 2007

IQ-Imbalance Compensation for OFDM in the Presence of IBI and Carrier-Frequency Offset
Imad Barhumi, Member, IEEE, and Marc Moonen, Member, IEEE
AbstractIn this paper, we propose a frequency-domain equalization technique for orthogonal frequency division multiplexing (OFDM) transmission over frequency-selective channels. We consider the case where the receiver analog front-end suffers from IQ-imbalance and the local oscillator suffers from carrier-frequency offset (CFO). While the IQ-imbalance results in a mirroring effect, the CFO induces inter-carrier interference (ICI). In addition to ICI, we consider the channel delay spread is larger than the cyclic prex (CP). This means that inter-block interference (IBI) is present. The frequency-domain equalizer is obtained by transferring a time-domain equalizer (TEQ) to the frequency-domain resulting in a per-tone equalizer (PTEQ). Due to the presence of IQ-imbalance the conventional TEQ (where only one TEQ is applied to the received sequence) is not sufcient to cope with the mirroring effect. A sufcient TEQ consists of two time-domain lters; one applied to the received sequence and another applied to a conjugated version of the received sequence. For the case of IQ-imbalance and CFO, the TEQs are designed according the basis expansion model (BEM) which showed to be able to cope with the ICI problem. Finally, in addition to the frequency-domain PTEQ design procedure, a training-based RLS type initialization scheme for direct per-tone equalization is also proposed. Index TermsBasis expansion model (BME), carrier-frequency offset (CFO), IQ-imbalance, orthogonal frequency-division multiplexing (OFDM), per-tone equalizer (PTEQ).

I. INTRODUCTION RTHOGONAL frequency-division multiplexing (OFDM) has been adopted for digital audio and video broadcasting [1] and chosen by the IEEE 802.11 standard [2], as well as by the HIPERLAN-2 standard [3] for wireless LAN (WLAN) transmission. This is due to its robustness against multipath fading and its relatively simple implementation. However,

Manuscript received June 13, 2005; revised January 16, 2006. This work was carried out at the ESAT Laboratory of the Katholieke Universiteit Leuven, in the frame of Belgian Programme on Interuniversity Attraction Poles, initiated by the Belgian Federal Science Policy Ofce IUAP P5/22 (Dynamical Systems and Control: Computation, Identication and Modelling) and P5/11 (Mobile multimedia communication systems and networks), the Concerted Research Action GOA-MEFISTO-666 (Mathematical Engineering for Information and Communication Systems Technology) of the Flemish Government, and Research Project FWO nr.G.0196.02 (Design of efcient communication techniques for wireless time-dispersive multi-user MIMO systems). The work of I. Barhumi was supported in part by the Palestinian European Academic Cooperation in Education (PEACE) Programme. The associate editor coordinating the review of this manuscript and approving it for publication was Dr. Kostas Berberidis. I. Barhumi is with the Faculty of Engineering, United Arab Emirates University, Al-Ain 17555, UAE (e-mail: imad.barhumi@uaeu.ac.ae). M. Moonen is with the Katholieke Universiteit (KU) Leuven, B-3001, Heverlee, Belgium (e-mail: marc.moonen@esat.kuleuven.be). Color versions of Figs. 46 available online at http://ieeexplore.ieee.org. Digital Object Identier 10.1109/TSP.2006.882102

OFDM is sensitive to analog front-end imperfections; mainly the amplitude- and phase-imbalance (IQ-imbalance) and carrier-frequency offset (CFO). In order to maintain orthogonality between subcarriers in OFDM systems, a cyclic prex (CP) that is equal to or longer than the channel impulse response length is required. This orthogonality is preserved providing that IQ-imbalance and CFO are not present, and the channel is time-invariant (TI) over the OFDM block period. In practice, it is very difcult to satisfy all of these conditions. The presence of IQ-imbalance and CFO and the fact that the channel impulse response length may be larger than the CP-length cause a severe degradation in performance for OFDM systems. This motivates us to search for alternative equalization techniques that are robust against these imperfections. In this paper, we consider OFDM transmission over TI channels with a CP-length that may be shorter than the channel impulse response length and the analog front-end suffers from IQ-imbalance as well as CFO. We assume that these analog front-end impairments are present at the receiver side, while the transmitter enjoys perfect analog front-end. Different approaches have been proposed to overcome the analog front-end problems for OFDM transmission. In [4] a training-based technique for CFO estimation is proposed assuming perfect IQ-balance. A maximum likelihood (ML) CFO estimation is proposed in [5], also assuming perfect IQ-balance. The IQ-imbalance only problem is treated in [6][8] assuming zero CFO. Joint compensation of IQ-imbalance and CFO is treated in [9] and [10]. In [9], it is assumed that CFO is corrected based on perfect knowledge of the IQ-imbalance parameters, and the IQ-imbalance parameters can be estimated correctly in the presence of CFO. The assumption here is valid for small CFO and small IQ-imbalance parameters, while these algorithms are incompetent to achieve the desired accuracy for moderate to large IQ-imbalance parameters and large CFO values. In [10], nulled subcarriers are used to estimate CFO by maximizing the energy on the designated subcarrier and its image. However, in the aforementioned works, the CP-length is consistently assumed to be longer than or equal to the channel impulse response length. In this paper, we consider the case where the CP-length may be shorter than the channel impulse response length, and furthermore the IQ-imbalance and CFO are present at the receiver side. We devise a frequency-domain per-tone equalization technique to combat the channel effect as well as the imperfection of the analog front-end. We rst treat the case of IQ-imbalance and zero CFO, and then extend the analysis to the case of IQ-imbalance and nonzero CFO. The per-tone equalizer (PTEQ) is obtained by transferring a time-domain equalizer (TEQ) to the frequency-domain. The purpose of the TEQ is to

1053-587X/$20.00 2006 IEEE

BARHUMI AND MOONEN: IQ-IMBALANCE COMPENSATION FOR OFDM

257

shorten the channel impulse response length to t within the CP-length and to eliminate the effect of the IQ-imbalance. The proposed PTEQ is rst designed assuming perfect knowledge of the channel as well as of the IQ-imbalance parameters and the CFO or assuming they can be accurately estimated. This is, however, far from being practical, since the channel estimate in the presence of IQ-imbalance and/or CFO is not sufciently accurate. To avoid the intermediate stage of parameters estimation, a training-based RLS-type initialization scheme is proposed for such per-tone equalization. Note that, knowing the IQ-imbalance and CFO perfectly may lead to a simpler compensation in the time-domain. The problem arises when estimating accurately the IQ-imbalance parameters in the presence of CFO, and CFO in the presence of IQ-imbalance. The same holds for estimating the channel in the presence of IQ-imbalance and CFO (either after or before compensation, or jointly). Which is hard to obtain with the necessary accuracy to estimate the transmitted symbols. The purpose of this paper is to develop a frequency-domain direct equalization/compensation scheme without passing through the estimation stage. This paper is organized as follows. In Section II, we introduce the system model. The per-tone equalizer for the IQ only case is introduced in Section III. In Section IV, the per-tone equalizer is extended to the case of IQ-imbalance and CFO. Simulation results are introduced in Section V. Finally, our conclusions are drawn in Section VI. Notation: We use upper (lower) bold face letters to denote matrices (column vectors). Superscripts , , and represent conjugate, transpose, and Hermitian, respectively. We denote and the Kronecker product as . We the expectation as identity matrix as , the all-zero denote the matrix as and all the ones vector of length as . The th element of vector is denoted by . We denote the set of positive integers without zero. Finally, denotes the diagonal matrix with vector on the diagonal. II. SYSTEM MODEL We consider OFDM transmission over time-invariant (TI) frequency-selective channels. We assume a single-input single-output (SISO) system, but the results can be easily extended to single-input multiple-output (SIMO) or multiple-input multiple-output (MIMO) systems. At the transmitter the information-bearing symbols are parsed into blocks of frequency-domain QAM symbols. Each block is then transformed to the time-domain by the inverse discrete Fourier transform (IDFT). A CP of length is added to the head of each block. The time domain blocks are then serially transmitted over the TI channel. In the ideal case where neither IQ-imbalance nor CFO are present, the discrete time-domain baseband equivalent description of the received signal at time index is given by

symbols per second. The sequence transmitted at a rate of channel is assumed to be time-invariant for a burst of OFDM symbols and may change independently from burst to burst. Asis the QAM symbol transmitted on the th subcarsuming can be written as rier of the th OFDM block,

where and . Note that this description includes the transmission of a CP of length . III. EQUALIZATION IN THE PRESENCE OF IQ-IMBALANCE Here, we tackle the problem of channel equalization for OFDM in the presence of IQ-imbalance and interblock interference (IBI), assuming zero CFO. Channel equalization for OFDM in the presence of IQ-imbalance and CFO is treated in Section IV. IQ-imbalance arises when the front-end does not preserve the power balance between the in-phase (I) and the quadrature (Q) branches of the direct downconversion receiver (referred to also as zero-IF). IQ-imbalance is characterized by and the phase two parameters: the amplitude-imbalance . In this paper, we focus on the problem of receiver mismatch IQ-imbalance while the transmitter enjoys perfect IQ-balance. Note that, due to the channel in-between, the IQ-imbalance at the receiver cannot be referred to as IQ-imbalance at the transmitter and vice versa. The discrete-time baseband received sequence in the presence of IQ-imbalance at time index can then be written as (1) where the parameters and are given by [11]

We further consider the CP length is smaller than the channel impulse response length, which means that IBI is present. To eliminate such IBI, time-domain ltering may be used to shorten the channel such that it ts within the CP length. Due to IQ-imbalance the received signal is a mixture of the desired signal and its image. In OFDM systems, this translates to the fact that the received signal at subcarrier is a composite of the desired signal at that subcarrier and an interfering image signal (referred to also as mirror signal). For from subcarrier this reason, the conventional time-domain shortening where one TEQ is applied to the received signal is not sufcient to cope with the problem in hand. However, it will be shown that two TEQs can be used, where one is applied to the received signal and the other is applied to a conjugated version of the received sequence, as shown in Fig. 1. The second TEQ is necessary here to compensate for the image symbol, as will become clear later. The output of the TEQs subject to some decision delay can be written as (2)

where is the th tap of the multipath fading channel, is the baseband equivalent noise, and is the discrete time-domain

258

IEEE TRANSACTIONS ON SIGNAL PROCESSING, VOL. 55, NO. 1, JANUARY 2007

Fig. 1. Time-domain equalizer block diagram.

On the block level formulation, (2) can be written as

be signicantly reduced by replacing the sliding DFT with one FFT and the remaining FFTs are compensated for by difference terms as explained in the following properties: (3) (6a) and (6b) where the frequency response of the received sequence on the th subcarrier is written as , and is an lower triangular Toeplitz matrix given by .. . . . with by .. . . .. .. . . . . .

where

, and (for ) is an Toeplitz and rst matrix with rst column . row In conjunction with the time-domain ltering, a one-tap frequency-domain equalizer (FEQ) is applied to the received sequence in the frequency-domain to recover the transmitted as the estimate of the transmitted QAM symbols. Dene QAM symbol on the th subcarrier of the th OFDM symbol. This estimate is then obtained as (4) where is the 1-tap FEQ operating on the th subcarrier, and is the st row of the DFT matrix . Transferring the TEQ operation to the frequency domain we obtain (5) where , and is given by

(7)

. The difference term vector

is given

. . .

. . . .. . .. . . . .

By dening (5) can be written as

, and

(8) Note that due to the IQ-imbalance the quantity contains information about as well as for . The same holds for the quantity . A proper equalizer should then consider the estimation of the transmitted QAM symbol on the th subcarrier and th subcarrier in a the transmitted QAM symbol on the

The estimate of the transmitted symbol on the th subcarrier of the th OFDM block is obtained by performing a sliding DFT on the received sequence, i.e., by performing an N-point DFT within a sliding-window of size . This corresponds to perN-point DFTs. The implementation of (5) can forming

BARHUMI AND MOONEN: IQ-IMBALANCE COMPENSATION FOR OFDM

259

Fig. 2. PTEQ for OFDM with IQ-imbalance.

joint fashion. The implementation of the PTEQ (8) is shown in Fig. 2. Hence, we can write the estimate of the transmitted QAM th subcarrier as symbol on the

where

is given by

(9)

with is an antidiagonal identity matrix of size . At this point, we may introduce a model for the received se. Dene quence , can be written as

Combining (8) and (9) we arrive at

(12) (10) where

The relationship between sequence can be drawn as

(dened in (10)) and the received

(11)

, and is an Toeplitz matrix with rst column and rst , is the CP insertion matrix row given by

260

IEEE TRANSACTIONS ON SIGNAL PROCESSING, VOL. 55, NO. 1, JANUARY 2007

and is the transmitted vector of is the QAM symbols transmitted in the th OFDM block. . Hence, we can write the noise vector similarly dened as received sequence and its conjugate as

and

For this case, the received symbol at the th subcarrier in the th can be written as OFDM block

(13)

(17) where is the channel frequency response at the th subis the noise on the th subcarrier in the th carrier and OFDM block. Hence, we can write (16) as

where

is an

matrix dened as (18) . . .

To obtain the PTEQ coefcients for the th and th ), we dene the following meansubcarriers (i.e., solve for square error (MSE) cost function

The minimum MSE (MMSE) solution can be then obtained as (14) The solution of (14) is obtained by solving is equal to and

(15) and are the transmitted sequence covariance mawhere is a trix, and the noise covariance matrix respectively, and long unity vector with 1 in the ( )th position. For the case of OFDM transmission over TI channels with a CP length that is longer than the channel impulse response and the decision length, we may consider a TEQ of order delay in this case need not necessarily to be different from zero ). Therefore, the estimate of in (10) becomes as (i.e., (16) with

In the noiseless case, perfect symbol estimation [i.e., zero-forcing (ZF) equalization] is possible as long as the channel does not experience deep fades on subcarrier or on subcarrier . Note that with only one TEQ, the ZF solution does not exist, while, the MMSE solution always exists. However, it can be easily shown that the MMSE solution with one TEQ has a very poor performance. Equation (15) requires that the channel and the IQ-imbalance parameters to be known at the receiver. If the IQ-imbalance parameters are known then probably it is better to do a time-domain IQ-compensation based on (1) [6], and then pertone equalization which leads to a much simpler solution. In practice, since it is hard to obtain the IQ-imbalance parameters and the channel impulse response sufciently accurate, we look for direct initialization schemes, which indeed skip the estimation stage. The previous analysis was essential to dene a proper equalizer structure. In the following we discuss a training-based recursive least squares (RLS) initialization scheme for direct equalization. RLS Initialization: In our discussion so far, we have assumed that the channel and the IQ-imbalance parameters are known at the receiver. In this subsection, we will derive a trainingbased RLS algorithm for direct equalizer design without passing through the intermediate stage of channel and IQ-imbalance parameters estimation. Assuming that OFDM symbols are dedfor the th subcarrier can be icated for training, the optimal computed as

(19) The problem in (19) can be solved recursively as listed in Algorithm 1, where the RLS algorithm is invoked.

BARHUMI AND MOONEN: IQ-IMBALANCE COMPENSATION FOR OFDM

261

Algorithm 1 RLS Direct Equalization for initialize initialize end for for for to do do do

to (3), the output of the TEQs in the th OFDM block can be written as (22) Unlike (3), the TEQs proposed in (23) are now time-varying. This is necessary in order to cope with the time-variation imposed on the system due to the presence of CFO. Therefore, the TEQs are assumed to be time-varying, and may change from block to block. The purpose of the TEQs is to eliminate IBI and ICI and to compensate for the IQ-imbalance and CFO. In other words, to restore orthogonality between subcarriers that was destroyed by IBI and/or ICI and at the same time to compensate for the mirroring effect induced by the IQ-imbalance. Hence, the TEQ is required to shorten the channel impulse response into a target impulse response whose length ts within the CP-length and eliminate ICI (eliminate the time-variations imposed on the system due to CFO). Designing and optimizing the performance of the TEQ is beyond the scope of this paper. In this paper, we are targeting frequency-domain per-tone equalizers, which are obtained by transferring the TEQ operation to the frequency-domain, as will become clear later. It was shown in [13] and [14] that modeling the TEQ using the basis expansion model (BEM) is an efcient way to tackle the problem of IBI/ICI for OFDM transmission over time-varying channels due to user mobility and carrier-freand can quency offset. Using the BEM, the TEQs be modeled as

end for end for

where

is a small positive constant.

IV. EQUALIZATION IN THE PRESENCE OF IQ-IMBALANCE AND CFO In Section III, we tackled the problem of per-tone equalization for OFDM in the presence of IQ-imbalance and zero CFO. In this section, we assume that in addition to the IQ-imbalance the receiver analog front-end suffers from CFO due to a drift in the local oscillator. As a result of CFO, orthogonality between subcarriers is destroyed which introduces ICI. In addition, we again assume that the CP length is smaller than the channel impulse response length, which introduces IBI. Assuming that the where local oscillator exhibits a normalized CFO of is the carrier frequency offset (in Hz). The baseband equivalent received sequence at time index is then given by [9] and [12] (20) On a block level formulation, (20) can be written as (21) , and . Similar to the case of IQ-imbalance only, we start from the time-domain equalization. In time-domain we apply two TEQs: one is applied to the received sequence and the other one is applied to a conjugated version of the received sequence. Similar where

(23)

where is a diagonal matrix with the th time-varying basis components of the the BEM on its diagonal , and (for ) is an Toeplitz matrix with rst and rst row column . In conjunction with the time-domain ltering, a 1-tap FEQ is applied in the frequency-domain to recover the transmitted QAM symbols. Hence, the estimate of the QAM symbol transmitted on the th subcarrier in the th OFDM block can be written as

(24)

262

IEEE TRANSACTIONS ON SIGNAL PROCESSING, VOL. 55, NO. 1, JANUARY 2007

Fig. 3. PTEQ for OFDM with IQ imbalance and CFO.

A PTEQ is then obtained by transferring the TEQ operation to the frequency-domain. Doing so we obtain the estimate of the transmitted QAM symbol at the th subcarrier in the th OFDM

Finally, dene can now be written as . . .

and

, (26)

(25)

. . .

(27)

where for . Using the properties introduced in (6a) and (6b) and dening and , (25) can now be written as

Due to the IQ-imbalance the mirroring effect arises, and hence a proper equalizer takes into account this effect by estimating the transmitted symbol on subcarrier and its mirror on subcarrier in a joint fashion. Writing a similar equation for the th subcarrier and combining it estimated symbol on the with (27) we arrive at: . . .

(26) . . . as

Dene

and . Dene the selection matrix

(28) where is the rst row of the matrix and rows of the matrix the matrix consisting of the last is . From (28), the estimate of the QAM transmitted symbols can be obtained by applying a time-varying PTEQ that varies from

BARHUMI AND MOONEN: IQ-IMBALANCE COMPENSATION FOR OFDM

263

block to block. However, the implementation/computation of the time-varying PTEQ can be signicantly reduced by splitting the PTEQ into two parts: a time-invariant part that is xed as long as the channel is time-invariant and a time-varying part that can be compensated for by means of a complex rotation applied to all subcarriers in the same OFDM block, as will become clear later on. dened in (28) can now be written as Note that

(32)

(29)

where

is given by

. . .

We can also write the received sequence and its conjugate as shown in (30) at the bottom of the page. , we dene the following MSE cost funcTo solve for tion:

Hence, the MMSE solution of can be obtained by setting , and obtained as

(31) For a white source , (31) can be written as

The MMSE PTEQ obtained in (32) for the case of a white input source consists of two parts: A time-invariant part denoted and a time-varying part captured by the complex rotaby . The time-invariant part is xed as long as the tion matrix channel, the IQ-imbalance parameters and the CFO are xed. The complex rotation, on the other hand, changes from block to block. Note that, the complex rotation within the same OFDM block is the same over all subcarriers. Exploiting this structure we can signicantly reduce the computational complexity by computing the time-invariant part of the PTEQ and compensating for the time-variations by means of the complex rotation afterwards. On the implementation level, the PTEQ is explained in Fig. 3. In the aforementioned discussion, we assume that the channel, the IQ-imbalance parameters and CFO are known at the receiver or they can be estimated with sufcient accuracy. In reality, it is difcult if not impossible to estimate the channel in the presence of IQ-imbalance and CFO, and or estimate the IQ-imbalance parameters in the presence of CFO or vice versa. Therefore, to bypass the estimation stage, direct equalization/compensation are favorable. In the following, we develop a training-based RLS direct equalization/compensation scheme. RLS Initialization: In order to obtain the PTEQ (32), the channel as well as the IQ parameters and the CFO must be known/estimated. In this section, we devise a training-based RLS direct equalization/estimation of the time-invariant part of the PTEQ assuming that only the CFO is known at the receiver. In a similar fashion to the RLS algorithm devised in Section III, assuming that OFDM symbols are available for training, the for the th subcarrier can be computed as optimal

(33) The RLS algorithm to compute the time-invariant part of the PTEQ is listed in Algorithm 2.

(30)

264

IEEE TRANSACTIONS ON SIGNAL PROCESSING, VOL. 55, NO. 1, JANUARY 2007

Algorithm 2 RLS Direct Equalization With CFO for initialize initialize end for for for to do do do

Fig. 4. IQ compensation.

end for end for

V. SIMULATIONS Here, we present some of the simulation results of the proposed equalization techniques for OFDM transmission over time-invariant channels. We consider an OFDM system subcarriers. The time-invariant channel is of with . The channel taps are simulated as i.i.d complex order Gaussian random variables with zero-mean and unit variance. The channel is kept xed over a burst of transmission, and may change from burst to burst independently. The analog front-end of the OFDM receiver suffers from IQ-imbalance and CFO. In the simulations uncoded quadrature phase-shift-keying (QPSK) signaling is assumed. The performance is measured in terms of bit-error-rate (BER) versus signal-to-noise ratio (SNR). The , where is the SNR is dened as SNR transmitted QPSK symbol energy, and is the noise power. First, we consider the case of IQ-imbalance only (i.e., zero CFO). We consider an OFDM receiver with phase imbalance and amplitude-imbalance . We focus on the frequency-domain PTEQ obtained by transferring a TEQ operation to the frequency domain. In the rst setup, we assume that perfect channel state inforand ) are mation (CSI) and IQ-imbalance parameters ( known at the receiver. In Fig. 4 we consider the following different scenarios. , and . This No IQ-imbalance, no IBI case corresponds to an ideal OFDM transmission over TI channels with perfect analog front-end. This case will serve as a benchmark for our equalizer. . The proposed PTEQ IQ-imbalance, no IBI and does not count for the IQ-imbalance (no compensation).

Fig. 5. IQ RLS direct equalization.

This case quanties the inuence of IQ-imbalance on system performance when no IBI is present. . When only the IQ-im IQ-imbalance, no IBI and balance is present and compensated for by means of . one-tap FEQ are present. To evaluate IQ-imbalance and IBI with the effect of the IQ-imbalance on system performance, we design an equalizer that takes into account the IBI but not the IQ-imbalance (no IQ-compensation). A PTEQ of order is considered. are present. We design an IQ-imbalance and IBI with equalizer that takes into account IBI and compensates for . the IQ-imbalance. The lter order is chosen to be IQ-imbalance and IBI with are present. We design an equalizer that takes into account IBI and compensates for the IQ-imbalance, but the lter is chosen now to be of . order As shown in Fig. 4, the IQ imbalance degrades the performance signicantly, where the BER curve suffers from an early error . Taking into account the IQ imbalance oor at enhances the system performance signicantly. A one-tap FEQ with IQ imbalance allows to approach the performance of ideal

BARHUMI AND MOONEN: IQ-IMBALANCE COMPENSATION FOR OFDM

265

Fig. 6. IQ and CFO direct equalization.

OFDM, when no IBI is present. When IBI is present, a PTEQ with IQ compensation allows to approach the of order performance of ideal OFDM for low to moderate SNR values, for high SNR values. and suffers from an error oor at reduces the error oor A similar PTEQ but of order . The error oor results here due to the fact that the to nite length PTEQ is incapable of completely eliminating ICI resulted due to IBI. In the second setup, we assume that the channel and the IQ imbalance parameters are unknown at the receiver. In this case the training-based RLS type direct equalization listed as Algorithm 1 is invoked. We consider an RLS scheme with and OFDM symbols are available for training/initialization. We compare the obtained results with the case of perfect knowledge of the channel and the IQ parameters. As shown in Fig. 5, the RLS scheme with experiences a 1 dB loss in SNR compared with the case when perfect CSI and IQ imbalance parameters are known at the receiver. On the other hand, the performance of the RLS scheme with coincides with the case when perfect CSI and IQ imbalance parameters are known at the receiver. Second, we consider that the OFDM receiver suffers from an IQ imbalance and CFO. The IQ imbalance parameters are assumed to be equal to the ones used before, i.e., IQ phase imbalance and IQ amplitude imbalance . The normalized CFO is assumed to be , where only the noninteger part of the normalized CFO is considered, since the integer part of the CFO results into a cyclic shift and does not contribute to the ICI. Again we focus on the frequency-domain PTEQ, obtained by transferring the TEQ operation to the frequency-domain. The PTEQ is designed according to the BEM

and the number of basis functions . of order The number of basis functions reects the span of the ICI that is considered in the equalization process (which is in this case 4 neighboring subcarriers on each side). In the rst setup, we consider that the channel, the IQ imbalance parameters and the CFO are known at the receiver. As shown in Fig. 6, the proposed PTEQ approaches the performance of ideal OFDM transmission over TI channels for low to moderate SNR values, while it suffers from an error oor at BER for high SNR values. The error oor arises due to the fact that the proposed PTEQ is incapable of completely eliminating ICI which results here due to CFO and IBI. Note also, that the error oor here is higher than the error oor appeared in the rst setup (see Fig. 4) due to the presence of CFO. In the second setup, only the CFO is assumed to be known at the receiver. The training-based RLS type direct equalization listed as Algorithm 2 is applied to obtain the time-invariant part of the PTEQ. For the RLS algorithm we consider that and OFDM symbols are available for training. As shown in Fig. 6, the RLS-based direct equalization suffers from an error oor at BER for . While the performance of RLS-based direct equalization with almost coincides with the performance when perfect CSI, IQ imbalance parameters and CFO are known at the receiver, which also suffer from an error oor at BER . However, the RLS-based direct equalization with slightly outperforms the case when perfect knowledge is assumed. This is due to the fact that for the case of perfect knowledge we approximate the noise covariance matrix , which is not the case even under the assumption of white noise due to the IQ imbalance. The RLS algorithm appears to utilize the exact noise covariance matrix.

266

IEEE TRANSACTIONS ON SIGNAL PROCESSING, VOL. 55, NO. 1, JANUARY 2007

VI. CONCLUSION In this paper, we have proposed a frequency-domain PTEQ for OFDM transmission over time-invariant channels with imperfect analog front-end receivers. The CP length is assumed to be smaller than the channel order, which means that IBI is also present. We have shown that analog front-end imperfections degrade the performance signicantly, and, therefore, equalization and analog front-end imperfection compensation are crucial. For the case of IQ imbalance only we have devised a PTEQ to compensate for the channel and IQ imbalance under the assumption of knowing the channel and the IQ imbalance parameters. A more realistic scenario is either by estimating these parameters or by obtaining the equalizer coefcients directly. In this sense, we have developed a training-based RLS initialization scheme. For the case of IQ imbalance and CFO, the resulting PTEQ is time-varying. Under the assumption of white input source, it was shown that the time-varying equalizer can be split into two parts: a time-invariant part, and a time-varying part captured by means of a complex rotation. In all cases, a training-based RLS scheme can be used to design the time-invariant part of the PTEQ. REFERENCES
[1] ETSI Digital Video Broadcasting (DVB); Framing Structure, Channel Coding and Modulation for Digital Terrestrial Television, ETSI Standard EN 300 744 v1.4.1, Aug. 2001. [2] Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specications: High-speed Physical Layer in the 5 GHz Band, IEEE 802.11 Task Group a, Part 11, 1999. [3] ETSI Broadband Radio Access Networks (BRAN): HIPERLAN Type 2: Physical (PHY) Layer, ETSI TS 101 475 V1.2.2 (200102), Aug. 1999. [4] P. H. Moose, A technique for orthogonal frequency division multiplexing frequency offset correction, IEEE Trans. Commun., vol. 42, pp. 29082914, Oct. 1994. [5] J. H. Yu and Y. T. Su, Pilot-assisted maximum-likelihood frequencyoffset estimation for OFDM systems, IEEE Trans. Commun., vol. 52, pp. 19972008, Nov. 2004. [6] J. Tubbax, B. Come, L. V. der Perre, L. Deneire, S. Donnay, and M. Engels, Compensation of IQ imbalance in OFDM systems, presented at the IEEE Int. Conf. Communications (ICC), Anchorage, AK, May 1115, 2003. [7] C.-L. Liu, Impacts of I/Q imbalance on QPSK-OFDM-QAM detection, IEEE Trans. Consum. Electron., vol. 44, pp. 984989, Aug. 1998. [8] A. Targhiat and A. H. Sayed, On the baseband compensation of IQ imbalance in OFDM systems, presented at the IEEE Int. Conf. Acoustics, Speech, Signal Processing (ICASSP), Montreal, QC, Canada, May 2004. [9] J. Tubbax, A. Fort, L. V. der Perre, S. Donnay, M. Engels, M. Moonen, and H. D. Man, Joint compensation of IQ imbalance and frequency offset in OFDM systems, presented at the IEEE Global Commun. Conf. (IEEE GLOBECOM), San Francisco, CA, Dec. 15, 2003.

[10] S. Fouladifard and H. Shaee, Frequency offset estimation in OFDM systems in presence of IQ imbalance, presented at the Int. Conf. Communication Systems, Singapore, Nov. 2002. [11] M. Valkama and V. Koivunen, Advanced methods for I/Q imbalance compensation in communication receivers, IEEE Trans. Signal Process., vol. 49, no. 10, pp. 23352344, Oct. 2001. [12] G. Xing, M. Shen, and H. Liu, Frequency offset and IQ imbalance compensation for OFDM direct conversion receivers, presented at the IEEE Int. Conf. Acoustics, Speech, Signal Processing (ICASSP), Hong Kong, 2003. [13] I. Barhumi, G. Leus, and M. Moonen, Per-tone equalization for OFDM over doubly selective channels, presented at the IEEE Int. Conf. Communications (ICC), Paris, France, Jun. 2004. [14] I. Barhumi, G. Leus, and M. Moonen, Time-domain and frequencydomain per-tone equalization for OFDM over doubly selective channels, Signal Process., vol. 84/11, pp. 20552066, 2004, Special Section Signal Processing in Communications. Imad Barhumi (M06) was born in Palestine in 1972. He received the B.Sc. degree in electrical engineering from Birzeit University, Birzeit, Palestine, in 1996, the M.Sc. degree in telecommunications from the University of Jordan, Amman, Jordan, in 1999, and the Ph.D. degree in electrical engineering from the Katholieke Universiteit (KU) Leuven, Leuven, Belgium, in 2005. From 1999 to 2000, he was with the Electrical Engineering Department, Birzeit University, as a Lecturer. After his Ph.D. graduation, he served one year as a Postdoctoral Research Fellow with the Electrical Engineering Department, KU Leuven. Currently, he is an Assistant Professor with the Electrical Engineering Department of the United Arab Emirates University, Al-Ain, UAE. His research interests are in the area of signal processing for mobile and wireless communications, in particular, channel estimation and equalization of rapidly time-varying mobile wireless channels.

Marc Moonen (M94) received the electrical engineering degree and the Ph.D. degree in applied sciences from the Katholieke Universiteit (KU) Leuven, Leuven, Belgium, in 1986 and 1990, respectively. Since 2000, he has been an Associate Professor with the Electrical Engineering Department, KU Leuven, where he is currently heading a research team of 16 Ph.D. postdoctoral candidates, working in the area of signal processing for digital communications, wireless communications, DSL, and audio signal processing. Dr. Moonen received the 1994 KU Leuven Research Council Award, the 1997 Alcatel Bell (Belgium) Award (with P. Vandaele), and was a 1997 Laureate of the Belgium Royal Academy of Science. He was chairman of the IEEE Benelux Signal Processing Chapter (19982002) and is currently a European Association for Signal, Speech and Image Processing (EURASIP) AdCom Member from 2000 to the present. He is the Editor-in-Chief for the EURASIP Journal on Applied Signal Processing (2003-present), and a member of the editorial board of Integration, the VLSI Journal, the IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS II (20022003), EURASIP Journal on Wireless Communications and Networking, and the IEEE SIGNAL PROCESSING MAGAZINE.