A Study of Channel Estimation Techniques Based on Pilot

Arrangement in OFDM Systems

Sinem Coleri, Mustafa Ergen, Anuj Puri
{csinem, ergen,anuj}@eecs.berkeley.edu
University of California at Berkeley
Abstract- The channel estimation
techniques for OFDM systems based
on pilot arrangement are
investigated. The channel estimation
based on comb type pilot
arrangement is studied through
different algorithms for both
estimating channel at pilot
frequencies and interpolating the
channel. The estimation of channel at
pilot frequencies is based on LS and
LMS while the channel interpolation
is linear interpolation, second order
interpolation, low-pass interpolation,
spline cubic interpolation, and time
domain interpolation. Time-domain
interpolation is obtained by passing
to time domain through IDFT, zero
padding and going back to frequency
domain through DFT. In addition,
the channel estimation based on
block type pilot arrangement is
performed by sending pilots at every
sub-channel and using this
estimation for a specific number of
following symbols. We have also
implemented decision feedback
equalizer for all sub-channels
followed by periodic block-type
pilots. We have compared the
performances of all schemes by
measuring bit error rate with
16QAM, QPSK, DQPSK and BPSK
as modulation schemes, and
multipath rayleigh fading and AR
based fading channels as channel
models.
I. INTRODUCTION
Orthogonal Frequency
Division Multiplexing (OFDM) has
recently been applied widely in
wireless communication systems due to
its high data rate transmission
capability with high bandwidth
efficiency and its robustness to
multipath delay. It has been used in
wireless LAN standards such as
American IEEE802.11a and the
European equivalent HIPERLAN/2 and
in multimedia wireless services such as
Japanese Multimedia Mobile Access
Communications.
A dynamic estimation of
channel is necessary before the
demodulation of OFDM signals since
the radio channel is frequency selective
and time-variant for wideband mobile
communication systems. The channel
estimation can be performed by either
inserting pilot tones into all of the
subcarriers of OFDM symbols with a
specific period or inserting pilot tones
into each OFDM symbol. The first one,
block type pilot channel estimation, has
been developed under the assumption
of slow fading channel. Even with
decision feedback equalizer, this
assumes that the channel transfer
function is not changing very rapidly.
The estimation of the channel for this
block-type pilot arrangement can be
based on LS or MMSE. The MMSE
estimate has been shown to give 10-15
dB gain in SNR for the same mean
square error of channel estimation over
LS estimate [1]. In [2], a low-rank
approximation is applied to linear
MMSE by using the frequency
correlation of the channel to eliminate
the major drawback of MMSE, which
is complexity. The later, the comb-type
pilot channel estimation has been
introduced to satisfy the need for
equalizing the significant changes even
in one OFDM block. The comb-type
pilot channel estimation consists of
algorithms to estimate the channel at
pilot frequencies and to interpolate the
channel.
The estimation of the channel
at the pilot frequencies for comb-type
based channel estimation can be based
on LS, MMSE or LMS. MMSE has
been shown to perform much better
than LS. In [3], the complexity of
MMSE is reduced by deriving an
optimal low-rank estimator with
singular-value decomposition.
The interpolation of the
channel for comb-type based channel
estimation can depend on linear
interpolation, second order
interpolation, low-pass interpolation,
spline cubic interpolation, and time
domain interpolation. In [3], second-
order interpolation has been shown to
perform better than the linear
interpolation. In [4], time-domain
interpolation has been proven to give
lower BER compared to linear
interpolation.
In this paper, our aim is to
compare the performance of all of the
above schemes by applying 16QAM,
QPSK, DQPSK and BPSK as
modulation schemes, and multipath
rayleigh fading and AR based fading
channels as channel models. In section
II, the description of the OFDM system
based on pilot channel estimation is
given. In section III, the estimation of
the channel based on block-type pilot
arrangement is discussed. In section IV,
the estimation of the channel at pilot
frequencies is presented. In section V,
the different interpolation techniques
are introduced. In section VI, the
simulation environment and results are
described. Section VII concludes the
paper.
Binary
Data
Output
Data
AWGN
w(n)
Map
16QAM/
QPSK
S/P
Pilot
Insertion
IDFT
Guard
Insertion
P/S
Channel
+ S/P
Guard
Removal
DFT
Channel
Estimation
P/S
Demap
16QAM/
QPSK
Fig. 1 Baseband OFDM system
X(k)
Y(k)
x(n)
y(n)
h(n)
x
f
(n)
y
f
(n)
II.SYSTEM DESCRIPTION
The OFDM system based on
pilot channel estimation is given in
Figure 1. The binary information is first
grouped and mapped according to the
modulation in signal mapper. After
inserting pilots either to all sub-carriers
with a specific period or uniformly
between the information data sequence,
IDFT block is used to transform the
data sequence of length N {X(k)} into
time domain signal {x(n)} with the
following equation:
( ) ( ) { ( )
( ) 1 1 , 2 , 1 , 0
2
1
0

N n
e k X k X IDFT n x
N
kn
j
N
k

where N is the DFT length.

Following IDFT block, guard
time, which is chosen to be larger than
the expected delay spread, is inserted to
prevent inter-symbol interference. This
guard time includes the cyclically
extended part of OFDM symbol in
order to eliminate inter-carrier
interference (ICI). The resultant OFDM
symbol is given as follows:
( )
( )
( )
( ) 2
1 , 1 , 0 ,
1 , , 1 , ,

'

+ +

N n n x
N N n n N x
n x
g g
f

where Ng is the length of the guard
interval.
After following D/A
converter, this signal will be sent from
the transmitter with the assumption of
the baseband system model. The
transmitted signal will pass through the
frequency selective time varying fading
channel with additive noise. The
received signal is given by:
( ) ( ) ( ) ( ) 3 n w n h n x y
f f
+
where w(n) is additive white gaussian
noise and h(n) is the channel impulse
response, which can be represented by:
[4]
( ) ( ) ( ) 4 1 0
1
0
2

N n
r
i
i
Tn f
N
j
i
Di
e h n h

where r is the total number of

propagation paths, hi is the complex
impulse response of the i
th
path, fDi is
the i
th
path Doppler frequency shift, is
delay spread index, T is the sample
period and i is the i
th
path delay
normalized by the sampling time.
At the receiver, after passing
to discrete domain through A/D and
low pass filter, guard time is removed:
( ) ( ) ( ) 5 1 , 1 , 0 + N n N n y n y
g f

Then y(n) is sent to DFT block
for the following operation:
( ) ( ) { ( )
( ) 6 1 , 2 , 1 , 0
1
2
1
0

N k
e n y
N
n y DFT k Y
N
kn
j
N
n

Assuming there is no ISI, [7]

shows the relation of the resulting Y(k)
to H(k)=DFT{h(n)}, I(k) that is ICI
because of Doppler frequency and
W(k)=DFT{w(n)}, with the following
equation:
( ) ( ) ( ) ( ) ( )
( ) 7 1 1 , 0
+ +
N k
k W k I k H k X k Y

Following DFT block, the

pilot signals are extracted and the
estimated channel He(k) for the data
sub-channels is obtained in channel
estimation block. Then the transmitted
data is estimated by:
( )
( )
( )
( ) 8 1 , 1 , 0 N k
k H
k Y
k X
e
e

Then the binary information
data is obtained back in signal
demapper block.
III. CHANNEL ESTIMATION
BASED ON BLOCK-TYPE
PILOT ARRANGEMENT
In block-type pilot based
channel estimation, the pilot is sent in
all sub-carriers with a specific period.
Assuming the channel is constant
during the block, it is insensitive to
frequency selectivity. Since the pilots
are sent at all carriers, there is no
interpolation error. The estimation can
be performed by using either LS or
MMSE [1] [2]. The LS estimate is
represented by:
{
( ) 9
, ,
1
0
1 1 0
1
]
]
]
]
]

N
N
LS
y
y
y
x x x diag X where
y X h

where xi is the pilot value sent at the i

th
subcarrier and yi is the value received at
the i
th
sub-carrier.
If the time domain channel
vector g is Gaussian and uncorrelated
with the channel noise, the frequency-
domain MMSE estimate of g is given
by:
( )
( ) ( ) ( )
( ) 10
1
2
1
1 N 1 N
N
W
0 1 N
N
W
1 N 0
N
W
00
N
W
k
N
n
j
nk
N
yy gy
MMSE
e
N
W
and F
where y R FR h

]
]
]
]

where Rgy and Ryy is cross covariance
matrix between g and y and the auto-
covariance matrix of y respectively.
When the channel is slow fading, the
channel estimation inside the block can
be updated using the decision feedback
equalizer at each sub-carrier. Decision
feedback equalizer for the k
th
sub-
carrier can be described as follows:
The channel response at the k
th
sub-carrier estimated from the
previous symbol {He(k)} is used to
find the estimated transmitted
signal {Xe(k)}.
( )
( )
( )
( ) 11 1 , 1 , 0 N k
k H
k Y
k X
e
e

{Xe(k)}is mapped to the binary
data through signal demapper
and then obtained back through
signal mapper as {Xpure(k)}.
The estimated channel {He(k)} is
updated by:
( )
( )
( )
( ) 12 1 , 0 N k
k X
k Y
k H
pure
e

Since the decision feedback
equalizer has to assume that the
decisions are correct, the fast fading
channel will cause the complete loss of
estimated channel parameters.
Therefore, as the channel fading
becomes faster, there happens to be a
compromise between the estimation
error due to the interpolation and the
error due to loss of channel tracking.
For fast fading channels, as will be
shown in simulations, the comb-type
based channel estimation performs
much better.
IV. CHANNEL ESTIMATION AT
PILOT FREQUENCIES IN
COMB-TYPE PILOT
ARRANGEMENT
In comb-type pilot based
channel estimation, the Np pilot signals
are uniformly inserted into X(k)
according to the following equation:
( ) ( )
( )
( ) 13
0 ,
1 , 1 , . inf

'

l m
p
x
L l data
l mL X k X

where L=number of carriers/Np and

xp(m) is the m
th
pilot carrier value.
We define {Hp(k) k=0,1,
Np}as the frequency response of the
channel at pilot sub-carriers. The
estimate of the channel at pilot sub-
carriers based on LS estimation is given
by:
( )
( )
( )
( ) 14 1 , 1 , 0
p
p
p
e
N k
k X
k Y
k H
where Yp(k) and Xp(k) are output and
input at the k
th
pilot sub-carrier
respectively.
Since LS estimate is
susceptible to noise and inter-carrier
interference (ICI), MMSE is proposed
while compromising complexity. Since
MMSE includes the matrix inversion at
each iteration, the simplified linear
MMSE estimator is suggested in [5]. In
this simplified version, the inverse is
only need to be calculated once. In [3],
the complexity is further reduced with a
low-rank approximation by using
singular value decomposition.
V. INTERPOLATION
TECHNIQUES
IN COMB-TYPE PILOT
ARRANGEMENT
In comb-type pilot based
channel estimation, an efficient
interpolation technique is necessary in
order to estimate channel at data sub-
carriers by using the channel
information at pilot sub-carriers.
The linear interpolation
method is shown to perform better than
the piecewise-constant interpolation in
[6]. The channel estimation at the data-
carrier k, mL<k<(m+1)L, using linear
interpolation is given by:
( ) ( )
( ) ( ) ( ) ( )
( ) 15 0
1
L l
m H
L
l
m H m H
l mL H k H
p p p
e e
<
+ +
+
The second-order interpolation
is shown to fit better than linear
interpolation[3]. The channel estimated
by second-order interpolation is given
by:
( ) ( )
( ) ( ) ( )
( )
( ) ( )
( )
( ) 16 /
1
1 0
1
1
,
2
1
1
, 1 1
0
,
2
1
1

'

+ +
+

N l
m
p
H c m
p
H c m
p
H c
c
c
c
where
l mL H k H
e e

The low-pass interpolation is
performed by inserting zeros into the
original sequence and then applying a
special lowpass FIR filter that allows
the original data to pass through
unchanged and interpolates between
such that the mean-square error
between the interpolated points and
their ideal values is minimized (interp
in MATLAB).
The spline cubic interpolation
produces a smooth and continuous
polynomial fitted to given data points
(spline in MATLAB).
IFFT
Gp(n)
FFT
Gp(n)
0
Fig. 2TimeDomainInterpolation
H
p
(k)
H(k)
The time domain interpolation
is a high-resolution interpolation based
on zero-padding and DFT/IDFT [7].
After obtaining the estimated channel
{Hp(k), k=0,1,Np-1}, we first convert
it to time domain by IDFT:
( ) ( ) ( ) 17 1 , 1 , 0 ,
1
0
2

p
N
k
N
kn
j
p N
N n e k H n G
p
p

Then, by using the basic

multi-rate signal processing properties
[8], the signal is interpolated by
transforming the Np points into N
points with the following method:
( )
( )
( )

'

< + +
<
<
18
1 1
2
, 1
1
2 2
, 0
1
2
0 ,
N n
p
N
N
p
N N n
p
G
p
N
N n
p
N
p
N
n n
p
G
N
G
The estimate of the channel at
all frequencies is obtained by:
( ) ( ) ( ) 19 1 0
1
0
2

N
n
nk
N
j
N
N k e n G k H

VI. SIMULATION
A) DESCRIPTION OF
SIMULATION
I. System parameters
OFDM system parameters
used in the simulation are as follows:
the number of sub-carriers is 1024,
pilot ratio is 1/8, guard length is 256
and carrier modulation is QPSK,
DQPSK, BPSK or 16QAM. We
assume to have perfect synchronization
since the aim is to observe channel
estimation performance. Moreover, we
have chosen the guard interval to be
greater than the maximum delay spread
in order to avoid inter-symbol
interference. Simulations are carried
out for different signal-to-noise (SNR)
ratios and for different Doppler
spreads.
II. Channel model
Two multipath fading channel
models are used in the simulations. The
1
st
channel model is the ATTC
(Advanced Television Technology
Center) and the Grande Alliance DTV
laboratorys ensemble E model, whose
static case impulse response is given
by:
( ) ( ) ( )
( ) ( ) ( ) ( ) 20 137 1 . 0 75 1 . 0 36 1296 . 0
17 1995 . 0 2 3162 . 0 ) (
+ + +
+ +
n n n
n n n n h


The 2
nd
channel model is the
simplified version of DVB-T channel
model, whose static impulse response is
given in Table I.
In the simulation, we have
used Rayleigh fading channel. In order
to see the effect of fading on block type
based and LMS based channel
estimation, we have also modeled
channel that is time-varying according
to the following autoregressive (AR)
model:
( ) ( ) ( ) ( ) 21 1 n w n ah n h + +
where a is the fading factor and w(k) is
AWGN noise vector. a is chosen to
be close to 1 in order to satisfy the
assumption that channel impulse
response does not change within one
OFDM symbol duration. In the
simulations, a changes from 0.90 to
1.
Table 1
Channel Impulse Response For Channel 2
Delay (OFDMsamples) Gain Phase(rad)
0 0.2478 -2.5649
1 0.1287 -2.1208
3 0.3088 0.3548
4 0.4252 0.4187
5 0.49 2.7201
7 0.0365 -1.4375
8 0.1197 1.1302
12 0.1948 -0.8092
17 0.4187 -0.1545
24 0.317 -2.2159
29 0.2055 2.8372
49 0.1846 2.8641
III. Channel estimation based on block-
type pilot arrangement
We have modeled two types of
block-type pilot based channel
estimation. Each block consists of a
fixed number of symbols, which is 30
in the simulation. Pilots are sent in all
the sub-carriers of the first symbol of
each block and channel estimation is
performed by using LS estimation.
According to the first model, the
channel estimated at the beginning of
the block is used for all the following
symbols of the block and according to
the second method, the decision
feedback equalizer, which is described
in section III, is used for the following
symbols in order to track the channel.
IV. Channel estimation based on comb-
type pilot arrangement
We have used both LS and
LMS to estimate the channel at pilot
frequencies. The LS estimator
description is given in section IV. The
LMS estimator uses one tap LMS
adaptive filter at each pilot frequency.
The first value is found directly through
LS and the following values are
calculated based on the previous
estimation and the current channel
output as shown in Figure 3.
LMS +
X
p
(k) e(k)
Y
p
(k)
Fig. 3. LMSScheme
-
After estimating the channel at
pilot frequencies by using either LS or
LMS, all of the possible interpolation
techniques (linear interpolation, second
order interpolation, low-pass
interpolation, spline cubic interpolation,
and time domain interpolation) are
applied to investigate the effects.

B) SIMULATION RESULTS
Figures 4,5,6 and 7 give the bit
error rate performance of channel
estimation algorithms for different
modulations and for rayleigh fading
channel, whose static channel response
is given in (20) with Doppler frequency
70Hz. In these simulations, Block-type
estimation showed 10-15dB higher
BER than that of comb-type estimation.
Fig. 4. BPSK (Channel 1) Rayleigh Fading
1.E-04
1.E-03
1.E-02
1.E-01
1.E+00
5 10 15 20 25 30 35 40
SNR
linear
second order
low-pass
spline
time domain
block
decision-feedback
LMS
Fig. 5. QPSK (Channel 1) Rayleigh Fading
1.E-03
1.E-02
1.E-01
1.E+00
5 10 15 20 25 30 35 40
SNR
linear
second order
low-pass
spline
time domain
block type
decision feedback
LMS
Fig. 6. 16QAM (Channel 1) Rayleigh Fading
1.E-03
1.E-02
1.E-01
1.E+00
5 10 15 20 25 30 35 40 SNR
linear
second-order
low-pass
spline
time domain
block type
decision feedback
LMS
Fig. 7. DQPSK (Channel 1) Rayleigh Fading
1.E-04
1.E-03
1.E-02
1.E-01
1.E+00
5 10 15 20 25 30 35 40 SNR
linear
second order
low-pass
spline
time domain
block type
decision feedback
LMS
The comb-type channel
estimation with low pass interpolation
achieves the best performance among
all the estimation techniques for BPSK,
QPSK and 16QAM modulation. The
performance among comb-type channel
estimation techniques usually ranges
from the best to the worst as follows:
low-pass, spline, time-domain, second-
order and linear. The result was
expected since the low-pass
interpolation used in simulation does
the interpolation such that the mean-
square error between the interpolated
points and their ideal values is
minimized. These results are also
consistent with those obtained in [3]
and [4].
Fig. 8. 16 QAM (Channel 1) AR f ading
1.E-03
1.E-02
1.E-01
1.E+00
5 10 20 30 SNR
block type a=0.9
block type a=0.99
LMS a=0.9
LMS a=0.99
direct f eedback a=0.9
direct f eedback a=0.99
block type a=0.999
LMS a=0.999
direct f eedback a=0.999
DQPSK modulation based
channel estimation shows almost the
same performance for all channel
estimation techniques except the comb-
type channel estimation with spline
interpolation method. The BER
performance for all estimation types is
much better than channel estimation
techniques with other modulations for
high SNR whereas it is worse for low
SNR.
The effect of fading on the
block type and LMS estimation can be
observed from Fig.8 for autoregressive
channel model with different fading
parameters. As the fading factor a in
equation (21) increases from 0.9 to
0.999, the performance of both block
based methods and LMS improves.
When fading is fast, this means higher
fading parameter, the estimation does
not improve as SNR increases. The
reason for this is that the tracking error
in fast fading channel avoids improving
the performance. On the other hand, for
slow fading channel, the BER of the
decision feedback block-type channel
estimation tracks the channel much
better compared to the other two
schemes as SNR increases.
Fig. 9. 16QAM (Channel 2) Rayleigh Fading
1.E-04
1.E-03
1.E-02
1.E-01
1.E+00
5 10 15 20 25 30 35 40
SNR
linear
second order
low-pass
spline
time domain
block type
decision f eedback
LMS
The general characteristics of
the channel estimation techniques
perform the same for Rayleigh fading
channel, whose static impulse response
is given in table 1 for 16QAM as can be
seen in Fig.9.
Fig. 10. 16QAM (Channel 1) Rayleigh Fading
1.E-03
1.E-02
1.E-01
1.E+00
10 20 30 40 50 60 70 80 90 100
Doppler freq(Hz)
linear
second order
low-pass
spline
time domain
block type
decision feedback
LMS
Figure 10 shows the
performance of channel estimation
methods for 16QAM modulation,
Rayleigh fading channel whose static
response is given in
(20) and 40dB SNR for different
Doppler frequencies. The general
behavior of the plots is that BER
increases as the Doppler spread
increases. The reason is the existence of
severe ICI caused by Doppler shifts.
Another observation from this plot is
that decision feedback block type
channel estimation performs better than
comb-type based channel estimation for
low Doppler frequencies as suggested
in [10] except low-pass and spline
interpolation. We also observe that
time-domain interpolation performance
improves compared to other
interpolation techniques as Doppler
frequency increases.
VII.CONCLUSION
In this paper, a full review of
block-type and comb-type pilot based
channel estimation is given. Channel
estimation based on block-type pilot
arrangement with or without decision
feedback equalizer is described.
Channel estimation based on comb-
type pilot arrangement is presented by
giving the channel estimation methods
at the pilot frequencies and the
interpolation of the channel at data
frequencies. The simulation results
show that comb-type pilot based
channel estimation with low-pass
interpolation performs the best among
all channel estimation algorithms. This
was expected since the comb-type pilot
arrangement allows the tracking of fast
fading channel and low-pass
interpolation does the interpolation
such that the mean-square error
between the interpolated points and
their ideal values is minimized. In
addition, for low Doppler frequencies,
the performance of decision feedback
estimation is observed to be slightly
worse than that of the best estimation.
Therefore, some performance
degradation can be tolerated for higher
data bit rate for low Doppler spread
channels although low-pass
interpolation comb-type channel
estimation is more robust for Doppler
frequency increase.

REFERENCES
[1] J.-J van de Beek, O. Edfors, M. Sandell,
S.K. Wilson and P.O. Borjesson,
On channel estimation in OFDM systems
in Proc. IEEE 45
th
Vehicular Technology
Conference, Chicago, IL, Jul. 1995, pp.
815-819
[2] O. Edfors, M. Sandell, J.-J van de Beek,
S.K. Wilson and P.O. Borjesson, OFDM
channel estimation by singular value
decomposition in Proc. IEEE 46
th
Vehicular Technology Conference, Atlanta,
GA, USA, Apr. 1996, pp. 923-927
[3] M. Hsieh and C. Wei, Channel estimation
for OFDM systems based on comb-type
pilot arrangement in frequency selective
fading channels in IEEE Transactions on
Consumer Electronics, vol. 44, no.1,
February 1998
[4] R. Steele, Mobile Radio Communications,
London, England, Pentech Press Limited,
1992
[5] U. Reimers, Digital video broadcasting,
IEEE Communications Magazine, vol. 36,
no. 6, pp. 104-110, June 1998.
[6] L. J. Cimini, Jr., Analysis and simulation
of a digital mobile channel using orthogonal
frequency division multiplexing, IEEE
Trans. Commun., vol.33, no. 7, pp. 665-
675, July 1985.
[7] Y. Zhao and A. Huang, A novel channel
estimation method for OFDM Mobile
Communications Systems based on pilot
signals and transform domain processing,
in Proc. IEEE 47
th
Vehicular Technology
Conference, Phoenix, USA, May 1997, pp.
2089-2093
[8] A. V. Oppenheim and R. W. Schafer,
Discrete-Time Signal Processing, New
Jersey, Prentice-Hall Inc., 1999
[9] Digital video broadcasting (DVB):
Framing, channel coding and modulation
for digital terrestrial television, Draft ETSI
EN300 744 V1.3.1 (2000-08).
[10] Y. Li, Pilot-Symbol-Aided Channel
Estimation for OFDM in Wireless
Systems, in IEEE Transactions on
Vehicular Technology, vol. 49, no.4, July
2000.