See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/230801448

Channel Estimation Algorithms for OFDM Systems

Conference Paper  in  International Journal of Signal and Imaging Systems Engineering · January 2011
DOI: 10.1504/IJSISE.2012.050324

CITATIONS READS
14 1,018

4 authors, including:

Prasanta kumar Pradhan Oliver Faust

Institute of Aeronautical Engineering Sheffield Hallam University
7 PUBLICATIONS   21 CITATIONS    115 PUBLICATIONS   2,186 CITATIONS   

SEE PROFILE SEE PROFILE

Sarat Kumar Patra

Indian Institute of Information Technology Vadodara
155 PUBLICATIONS   657 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Hardware implementation of Fuzzy Logic Controllers View project

Design and Analysis of Bow-tie Antennas for GPR Applications View project

All content following this page was uploaded by Sarat Kumar Patra on 21 May 2014.

The user has requested enhancement of the downloaded file.

Channel Estimation Algorithms for OFDM Systems
Prasanta Kumar Pradhan∗ , Oliver Faust† , Sarat Kumar Patra∗ and Beng Koon Chua†
∗ Dept. of Electrons & Comm. Engg.
National Institute of Technology, Rourkela, India 769008
Email: pradhanp, [email protected]
† DSP Center, School of Engineering

Ngee Ann Polytechnic, Singapore 599489

Email: fol2, [email protected]

Abstract—In this work we have compared different type of function for the previous OFDM data block is used as the
channel estimation algorithm for Orthogonal Frequency Division transfer function for the present data block. In practice, the
Multiplexing (OFDM) systems. The result of the Adaptive Boost- channel transfer function of a wideband radio channel may
ing (AdaBoost) algorithm was compared with other algorithms
such Least Square (LS), Best Linear Unbiased Estimator (BLUE) have significant changes even within one OFDM data block.
and Minimum Mean Square Error (MMSE). Therefore, it is preferable to estimate channel characteristic
Index Terms—OFDM, Channel Estimation, LS, MMSE. based on the pilot signals in each individual OFDM data block
[3].
I. I NTRODUCTION Recently, an elegant channel estimation method for OFDM
In communication system many techniques, like Frequency mobile communication systems has been proposed by [4] in
Division Multiplexing Access (FDMA), Time Division mul- which a semi-blind low complexity frequency domain based
tiplexing Access (TDMA) and Code Division Multiplexing channel estimation algorithm for multi-access OFDM systems
Access (CDMA), are used for transmission of signal. Where was proposed [4]. Many researchers have pursued channel
FDMA has very bad spectrum usage and TDMA performance estimation in the time domain. A joint carrier frequency
degrades by multipath delay spread causing Inter Symbol synchronization and channel estimation scheme, using the
Interference (ISI). In contrast OFDM enables high bit-rate expectation-maximization (EM) approach, is presented in [5]
wireless applications in a multipath radio environment the need while [6] used subspace tracking. In [7], a joint and data
for complex receivers. OFDM is a multi-channel modulation estimation algorithm is presented which makes a collective use
system employing Frequency Division Multiplexing (FDM) of data and channel constraints. A joint frequency-offset and
of orthogonal sub-carriers, each modulating a low bit-rate channel estimation technique for multi-symbol encapsulated
digital stream. OFDM uses N overlapping (but orthogonal) (MSE) OFDM system is proposed in [8], while the authors of
sub bands, each carrying a baud rate of 1/T and they are [9] presented a sequential method based on carrier frequency
spaced 1/T apart. Because of the selected frequency spacing, offset and symbol timing estimation. The authors of [10]
the sub-carriers are all mathematically orthogonal to each estimated the channel based on power spectral density (PSD)
other. This permits proper demodulation of symbol streams and least squares (LS) estimation for OFDM systems with
without the requirement of non overlapping spectra. timing offsets. A pilot aided channel estimation algorithm
Currently, there is increasing interest in OFDM as it com- in the presence of synchronous noise by exploiting the a
bines the advantages of high data rates and easy implementa- priori available information about the interference structure
tion. This is reflected by the many standards that considered was presented in [11] while [12] used implicit pilots for joint
and adopted OFDM those for Digital Audio Broadcast (DAB) detection and channel estimation. A joint time domain tracking
and Digital video Broadcast (DVB), high speed modems over of channel frequency offset for OFDM systems is suggested
digital subscriber lines, and Wireless Local Area Network in [13] while a time domain carrier frequency offset (CFO)
(WLAN) broadband systems as of ieee 802.11a, 802.11b and tracking method based on Particle filtering is presented in
802.11g. [14]. Radial Basis Function (RBF) network based channel
The accuracy of channel state information greatly influences estimation has been investigated in [15]. This is based on
the overall system performance [1]. The main challenges asso- using the radial basis function (RBF) network to model the
ciated with OFDM systems today are- channel identification dynamics of the fading process. In one-dimensional channel
and tracking, channel coding and equalization. In wideband estimation, only the time-correlation of the fading channel
mobile channels, pilot-based signal correction schemes are is exploited, whereas in two-dimensional channel estimation,
feasible method for OFDM systems [2]. Most channel esti- both the time and the frequency correlations of the fading
mation methods for OFDM transmission systems have been channels are exploited. RBF networks are essentially nonlinear
developed under the assumption of a slow fading channel, interpolators of the pilot channels. AdaBoost based channel
where the channel transfer function is assumed stationary estimation is proposed in [16]. AdaBoost not only increases
within one OFDM data block. In addition, the channel transfer the performance of the OFDM systems it also renders the
 Bit Serial Add Parallel
Pilot Pilot Carrier
Stream Mapping To IFFT Cyclic To

b
Data Carrier Pilot Carrier
Insert

b
Prefix Serial Data Carrier

bc
Parallel

bc
b b b b b b b b b b b b b b
b b b b b b b b b b b b b b
bc bc bc bc bc bc bc bc bc bc bc bc bc bc
bc bc bc bc bc bc bc bc bc bc bc bc bc bc
Channel bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc
bc bc bc bc bc bc bc bc bc bc bc bc bc bc

Frequency
bc bc bc bc bc bc bc bc bc bc bc bc bc bc
bc bc bc bc bc bc bc bc bc bc bc bc bc bc

Time
bc bc bc bc bc bc bc bc bc bc bc bc bc bc

⊕
b b b
AWGN
b b b b b b b b b b b b b b b b b b b b b b b b b
bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc
bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc
bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc bc
Desired
bc bc
bc bc bc bc bc bc bc bc bc bc bc bc bc bc
Bit
bc bc bc bc bc bc bc bc bc bc bc bc bc bc
Remove Serial
stream Parallel Pilot
Demap FFT Cyclic To Frequency Time
To Extract
Serial Prefix Parallel
(a) Block Type (b) Comb Type

Channel Fig. 2. Pilot Arrangements

Estimate

Where r the total number of propagation paths, hi is the

Fig. 1. Block Diagram of the OFDM systems complex impulse response of the ith path, fDi is the ith
path’s Doppler frequency shift which causes Inter channel
Interference (ICI) of the received signals, λ is the delay spread
QAM mapping obsolete and thereby reducing the complexity index, and τi is the ith path delay time normalized by the
of receiver designs. sampling time. After removing the guard interval from yg (n),
The rest of the article is arranged as follows: Section. 2 the received samples y(n) are sent to a Fast Fourier Transform
describes the mathematical model of OFDM systems and its (FFT) block to demultiplex the multi-carrier signals.
channel model along with two types of pilot arrangement. The PN −1
Y (k) = F F T {y(n)} = N1 n=0 y(n)e−j2πkn/N (5)
implementation of AdaBoost and other algorithms (like LS,
MMSE, and BLUE) is discussed in Section. 3 . Section. 4 f or k = 0, 1, . . . , N − 1
describes the performance analysis of different algorithms and If we assume that the guard interval is longer than the length
finally Section. 5 draws up the conclusions. of channe1 impulse response- there is no inter-symbol inter-
II. S YSTEM D ESCRIPTION ference between OFDM symbols- the demultiplexed samples
Y (k) can be represented by

The block diagram of an OFDM system is given in Figure.1. Y (k) = X(k)H(k) + W (k), k = 0, 1, . . . , N − 1 (6)
The binary information data are grouped and mapped into j2πf T
sin(πfDi T ) 2πτi
Where H(k) = hi e Di
e−j N k and W (k) is πfDi T
multi-amplitude-multi-phase signals. In this paper, we consider
the Fourier transform of the AWGN w(k).
the 16-QAM modulation. After pilot insertion, the modulated
After that, the received pilot signals {Yp (k)} are extracted
data X(k) are sent to an IDFT and are transformed and
from {Y (k)}, the channel transfer function {H(k)} can be
multiplexed into x(n) as
obtained from the information carried by {Hp (k)}. With the
knowledge of the channel responses {H(k)}, the transmitted
PN −1
x(n) = IF F T {X(k)} = k=0 X(k)e
j2πkn/N
(1)
f or n = 0, 1, ..., N − 1 data samples {X(k)} can be recovered by simply dividing the
received signal by the channel response:
where N is the number of subcarriers. The guard interval
y(k)
Ng is inserted to prevent possible inter-symbol interference x̂(k) = (7)
in OFDM systems, and the resultant samples xg (n) are Ĥ(k)
where Ĥ(k) is an estimate of H(k) . After signal demapping,

x(N + n) n = Ng , Ng − 1, . . . , −1
xg (n) = (2) the source binary information data are reconstructed at the
x(n) n = 0, 1, . . . , N − 1
receiver output.
where Ng is the number of samples in the guard interval. The
The OFDM transmission scheme makes it easy to assign
transmitted signal is then sent to a frequency selective multi-
pilots in both time and frequency domain. Figure.2 shows
path fading channel. The received signal can be represented
two major types of pilot arrangement. The first kind of pilot
by
arrangement shown in Figure.2a is denoted as block-type pilot
yg (n) = xg (n) ⊗ h(n) + w(n) (3)
arrangement. The pilot signal is assigned to a particular OFDM
Where h(n) is the channel impulse response (CIR) and w(n) block, which is sent periodically in time domain. This type
is the Additive White Gaussian Noise (AWGN) and ⊗ is the of pilot arrangement is especially suitable for slow-fading
circular convolution. The channel impulse response h(n) can radio channels. The estimation of channel response is usually
be expressed as [17] obtained by either LS or MMSE estimates of training pilots
r−1 [5].
n
The second kind of pilot arrangement, shown in Figure. 2b,
X
h(n) = hi ej2πfDi T N δ(λ − τi ) (4)
i=0 is denoted as comb-type pilot arrangement. The pilot signals
are uniformly distributed within each OFDM block. Assuming systems which are based on comb-type pilot arrangement is
that the payloads of pilot signals of the two arrangements highly dependent on the rigorousness of estimate of pilot sig-
are the same, the comb-type pilot assignment has a higher nals. Thus, a estimate better than the LS estimate is required.
retransmission rate. Thus, the comb-type pilot arrangement The MMSE estimate has been shown to be better than the LS
system is provides better resistance to fast-fading channels. estimate for channel estimation in OFDM systems based on
Since only some sub-carriers contain the pilot signal, the block-type pilot arrangement. The Mean Square Error (MSE)
channel response of nonpilot subcarriers will be estimated by estimation, given in [17], shows that MMSE estimate has about
interpolating neighboring pilot sub-channels. Thus, the comb- 10-15 dB gain in SNR over the LS estimate for the same MSE
type pilot arrangement is sensitive to frequency selectivity values. The major drawback of the MMSE estimate is its high
when comparing to the block-type pilot arrangement system. complexity, which grows exponentially with the observation
That is, the pilot spacing (∆f )p , must be much smaller than samples.
the coherence bandwidth of the channel (∆f )c [3].
B. MMSE Estimation
III. C HANNEL E STIMATION The MMSE estimator employs the second order statistics
For comb-type pilot sub-carrier arrangement, the Np pilot of the channel conditions in order to minimizes Mean Square
signals Xp (m) , m = 0, 1, ..., Np − 1, are uniformly inserted Error (MSE). Let Rhh , RHH , and RY Y be the autocovariance
into X(k). That is, the total N sub-carriers are divided into Np matrix of h, H, and Y , respectively and RhY be the cross
groups, each with L = N/Np adjacent sub-carriers. In each covariance matrix between h and Y . Also σw 2
denotes the
group, the first sub-carrier is used to transmit pilot signal. AWGN variance E{|w |}. Assume the channel vector h, and
2

The OFDM signal modulated on the kth sub-carrier can be the AWGN w are uncorrelated, it can be derived that [18]
expressed as
RHH = E{HH H } = E{(F h)(F h)H } = F Rhh F H (13)

X(k) = X(mL + l) where

Xp (m), l = 0,
=
 
0(N −1)
inf ormation data, l = 1, 2, . . . , L − 1 T WN00 ... T WN
(8)  .. .. .. 
(14)
F = . . .

Let 
(N −1)0 (N −1)(N −1)

T WN ... T WN
HP = [Hp (0) Hp (1) . . . Hp (Np − 1)]T
= [H(0) H(L − 1) . . . H((Np − 1)(L − 1))]T k
and T WNi,k = √1N e−j2πi N is called as the twidle factor
(9)
matrix. Assuming Rhh (thus RHH ) and σw 2
are known at the
be the channel response of pilot carriers, and
receiver in advance, the MMSE estimator of the h is given by
Yp = [Yp (0) Yp (1) . . . Yp (Np − 1)] (10) ĥM M SE = RhY RY Y Y Y H . At last it can be estimated that

be a vector of received pilot signals. The received pilot signal ĤM M SE = F ĥM M SE = RHH [RHH +σw 2
(XX Ht )−1 ]−1 ĤLS
vector Yp can be expressed as (15)
MMSE employs the second order statistics of the channel
Yp = Xp Hp + Wp (11) for estimation. Some times the channel statistics are not avail-
where  able, so it is difficult to estimate the channel in this situation.
However, in OFDM systems the signal can be available at the

Xp (0) 0 ... 0
Xp =  ... ..
.
receiver by means of pilot carriers.
 
Xp (1) . . . 
0 0 . . . Xp (Np − 1) C. BLUE
where Wp is the vector of Gaussian noise in pilot sub-carriers.
If we restrict the estimator to be linear in data and and find
A. LS Estimation the linear estimator that is unbiased and minimum variance
In conventional comb-type pilot based channel estimation then the estimator is called Best Linear Unbiased Estimator
methods, the estimation of pilot signals, is based on the LS (BLUE). BLUE can be determined with knowledge of only
method is given by the first and second moment of the PDF. Since complete
knowledge of the PDF is not necessary, the BLUE is more
Ĥp,ls = [Hp,ls (0) Hp,ls (1) . . . Hp,ls (Np − 1)] suitable for practical implementation [19].
= Xp−1 Yp (12)
Y (0) Y (1) Y (N −1) Gauss-Markov Theorem 1. If the data can be modeled in
= [ Xpp (0) Xpp (1) . . . Xpp (Npp −1) ] the following linear form
The LS estimate of Hp is susceptible to AWGN and Inter- Y = XH + W (16)
Carrier Interference (ICI). Because the channel responses of
data subcarriers are obtained by interpolating the channel where X is a known N × p matrix and H is a p × 1 vector of
characterstics of pilot subcarriers, the performance of OFDM parameters to be estimated, and W is a N × 1 noise vector
 TABLE I
with zero mean and covariance C (the PDF of W is otherwise OFDM PARAMETERS
arbitrary ), then BLUE of H is given by
No of Subcarriers 256
HBLU E = (H Ht C −1 H)−1 H Ht C −1 X (17) No. of Pilot Carriers 32
Guard Interval 64
where C = (X − E{x})(X − E{x})Ht and Ht is the con- Guard Type Cyclic Extension
jugate transpose or Hermitian Transpose. and the minimum
variance of Ĥi is
var(Ĥi ) = [(H Ht C −1 H)−1 ]ii (18)
D. AdaBoost
Boosting is a general method for improving the accuracy of
any given learning algorithm. AdaBoost was originally defined
for two class problems but it can be extended boosting to
multi-class and regression problems. The AdaBoost algorithm,
introduced in 1995 by Freund and Schapire, solved many of
the practical problems of the earlier boosting algorithms [20].
The AdaBoost algorithm for multiclass problem is described
as below.
Suppose we are given a set of training data
(x1 , c1 ), (x2 , c2 ), . . . , (xn , cn ), where the input (prediction
variable) xi ∈ Rp and the output (response variable)
ci is quantitative values in afinite set, e.g. 1, 2, . . . , K.
Where K is the number of classes. Usually it is assumed
that the training data are composed from independently Fig. 3. Performance Comparison of Different Algorithms
and identically distributed (iid) samples from a unknown
probability distribution P rob(X, C). The gaol is to find out a
classification rule C(x) from the training data, so that when a) Fit a classifier T (m) (x) to the training data using
given a new input x, we can assign a class label c from weights wi .
1, 2, . . . , K. The misclassification error rate of a classifier b) Compute
PK
C(x) is given by 1 − k=1 EX [IC(X)=k P rob(C = k|X)]. n
X n
X
So from [20] err(m) = wi I(ci 6= T (m) (xi ))/ wi
i=1 i=1
C ? (x) = arg maxk P rob(C = k|X = x) (19) (20)
c) Compute
will minimize this quantity with the misclassification error rate
equal to 1 − EX maxk P rob(C = k|X). This classifier is 1 − err(m)
α(m) = log + log(K − 1) (21)
known as Bayes Classifier and its rate is Bayes Error Rate. err(m)
The multiclass AdaBoost algorithm tries to approximate the d) Set
Bayes Classifier C ? (x) by combining many weak classifiers.
Starting with an unweighted training sample, the AdaBoost wi ← wi exp(α(m) . I(ci 6= T (m) (xi ))) (22)
builds a classifier that produces class labels. If a training data for i=1, 2, . . . , n.
point is misclassified, the weight of that training data point e) Re-normalize wi .
is increased (boosted). A second classifier is built using these
3) Output
new weights, which are no longer equal. Again, misclassified
training data have their weights boosted and the procedure M
X
is repeated. Typically, one may build 500 or 1000 classifiers C(x) = arg maxk α(m) I (T (m) (x) = k)
this way. A score is assigned to each classifier, and the final m=1
(23)
classifier is defined as the score weighted linear combination
of the classifiers from each stage. Specifically, let T (x) denote IV. S IMULATION R ESULTS
a weak multi-class classifier that assigns a class label to x, then
the AdaBoost algorithm proceeds as follows: [21]
The OFDM system channel estimation was simulated with LS,
Multiclass AdaBoost Algorithm: 1. The Algorithm is as MMSE, BLUE and AdaBoost methods. In all simulations we
follows: have used QAM16 as the modulation scheme for the individual
1) Initialize the observation weights wi = 1/n, i = carriers. Other parameters of the simulation are given in the
1, 2, . . . , n. Table.I. Figure.3 shows comparison of Bit Error Rate (BER)
2) For m = 1 to M: for LS, MMSE, BLUE and AdaBoost. Figure.3 shows that
the AdaBoost algorithm improves the performance specially [14] T.Nyblom, T.Roman, M.Enescu, and V.Koivunen, “Time-varying carrier
in low SNR values. However, at high SNR both MMSE offset tracking in ofdm systems using particle filtering,” in Proceedings
of the Fourth IEEE International Symposium on Signal Processing and
and AdaBoost show a similar performance. Furtheremore, Information Technology, 2004, pp. 217–220.
AdaBoost gives better performance when compared to LS and [15] X. Zhou and X. Wang, “Channel estimation for ofdm systems using
BLUE. Nevertheless, BLUE and LS are performing same or adaptive radial basis function networks,” IEEE TRANSACTIONS ON
VEHICULAR TECHNOLOGY, vol. 52, no. 1, pp. 48–59, January 2003.
we can tell BLUE gives very marginal improvement to LS. [16] P. K. Pradhan, O. Faust, S. K. Patra, and B. K. Chua, “Channel estima-
The reason for this performance increase is because of the tion of ofdm system using adaptive boosting algorithm,” in Submitted
covariance matrix used in the BLUE. As our noise is ASWGN to ICC 2011.
[17] J. V. de Beek, O. Edfors, M. Sandel, S. Wilson, and P. Borjesson, “On
and it has variance of 1 so the BLUE’s performance is all most channel estimation in ofdm systems,” in IEEE 45th Vehicular Technology
that of LS algorithm. Conference, 1995.
[18] Y. Shen and E. Martinez, “Channel estimation in ofdm systems,” in
V. C ONCLUSSION Frescale Semiconductor Application Note, 2006.
[19] S. M. Kay, Fundamentals of Statistical Signal Processing- Estimation
The computational complexity of the MMSE increases Theory. Prentice Hall PTR, 1993, vol. 1.
exponentially as the number of carrier increases. Whereas [20] Y. Freund and R. E. Scapire, “A short introduction to boosting,” Journal
the computatinal complexity is not exponential in the case of Japaneese Society for Artificial Intelligence, vol. 14, no. 5, pp. 771–
780, 1999.
of AdaBoost. Hence, AdaBoost can be employed when a [21] J. Zhu, H. Z. S. Rosset, and T. Hastie, “Multiclass adaboost,” Journal
high number of carriers is required. Furthermore, as it is a of Statistics and it’s Interface, vol. 2, pp. 349–360, 2009.
classification algorithm so in the receiver side we will require
a separate demapper (or decoder) to get the desired data bits.
AdaBoost not only increases the performance of the OFDM
systems it also renders the QAM mapping obsolete and thereby
reducing the complexity of receiver designs.
R EFERENCES
[1] Y. Li and G. Stber, Orthogonal Frequency Division Mutiplexing for
Wireless Communication, Y. Li and G. Stber, Eds. Springer, 2006.
[2] F. Tufvesson and T. Masseng, “Pilot aided channel estimation for ofdm
in mobile cellular system,” in in Proceeding IEEE 47th Vehicular
Technology Conference. Phoneix, USA: IEEE, May 1997, pp. 11 639–
1643.
[3] M.-H. Hsieh and C.-H. Wei, “Channel estimation for ofdm systems
based on comb type pilot arrangement in frequency selective fading
channels,” IEEE Transactions on Consumer Electronics, vol. 44, no. 1,
pp. 217–225, February 1998.
[4] M. Sohail and T.Y.Al-Naffouri, “An em based frequency domain channel
estimation algorithm for multi access ofdm systems,” Elsevier Journal
of Signal Processing, vol. 10, pp. 1562–1572, 2010.
[5] J. Lee, J. Han, and S. Kim, “Joint carrier frequency synchronization
and channel estimation for ofdm systems via em algorithm,” IEEE
Transactions on Vehicular Technology, vol. 55, no. 1, pp. 167–172, 2006.
[6] X. Hou, Z. Zhang, and H. Kayama, “Time domain mmse channel
estimation based on sbspace tracking for ofdm systems,” in IEEE 63rd
Vehicular Technology Conference, Melbourne, Victoria, May 2006, pp.
1565–1569.
[7] T. Al-Naffouri, “An em based forward backward kalman filter for
estimation time variant channels in ofdm,” IEEE Transaction on Signal
Processing, vol. 1, no. 11, pp. 1–7, 2007.
[8] X. Wang, Y. Wu, J. Chouinard, and H. Wu, “On the design and
performance analysis of multisymbol encapsulated ofdm systems,” IEEE
Transactions on Vehicular Technology, vol. 55, no. 3, pp. 990–1002,
2006.
[9] J. Liu and J. Li, “Parameter estimation and error reduction for ofdm
based wlans,” IEEE Transactions on Mobile Computing, vol. 3, no. 2,
pp. 152–163, 2004.
[10] S. Qin, P. Liu, L. Zheng, and D. Wang, “Channel estimation with timing
offset based on psd & ls estimation for wireless ofdm systems,” in
International Symposium on Intelligent Signal Processing and Commu-
nication Systems, November-December 2007, pp. 248–251.
[11] A. Jeremic, T. Thomas, and A. Nehorai, “Ofdm channel estimation in
the presence of interference,” IEEE Transactions on Signal Processing,
vol. 52, no. 12, pp. 3429–3439, 2004.
[12] R.Dinis, N.Souto, J.Silva, A.Kumar, and A.Correia, “Joint detection and
channel estimation for ofdm signals with implicit pilots,” in IEEE Mobile
and Wireless Communications Summit, July 2007, pp. 1–5.
[13] T.Roman, M.Enescu, and V.Koivunen, “Joint time-domain tracking of
channel and frequency offsets for mimo ofdm systems,” in Wireless
Personal Communications, vol. 31, no. 3-4, 2004, pp. 181–200.

View publication stats