Novel Weigted

Proceedings of IC-NIDC2010
A NOVEL WEIGHTED CHANNEL ESTIMATION

ALGORITHM WITH NOISE REDUCTION FOR
OFDM SYSTEM
Xinjun Shi, Tao Luo, Miao Hu, Changchuan Yin
Key Laboratory of Universal Wireless Communications, Ministry of Education,
Beijing University of Posts and Telecommunications, Beijing, 100876, China
[email protected], [email protected], [email protected], [email protected]
Abstract
In this paper, we propose a p-weighted noise
reduction algorithm to improve the accuracy of
channel estimation in Orthogonal Frequency
Division Multiplexing (OFDM) systems. A SNR
and speed adaptive weighting factor is
introduced, which can efficiently reduce the noise.
After that, an exact expression of the optimal value
of under piecewise linear interpolation in time
domain are derived, as well as a more practical
near-optimization of . Simulation results show
that p-weighted algorithm obtains a SNR gain at
low SNR and maintains the variation of channel at
high SNR. It achieves the optimal tradeoff between
noise reduction and maintenance of channel
variation.
Keywords: OFDM ; time-varying channel
noise reduction; channel estimation
1 Introduction
OFDM is widely used in wireless communication
systems which usually suffer from time and
frequency fading channels. Accurate estimation of
the wireless channel is crucial for improving the
performance of OFDM systems. Channel
estimation, in general, is carried out by pilot
estimation and then channel coefficient
interpolation. As the simplest method, least square
(LS) estimation is always utilized to initialize
channel estimates on the pilot position [1]. After
that, other methods such as interpolating will be
used to get all estimates.
As regards interpolation, piecewise linear method
[2] is the simplest one, which however is sensitive
to AWGN. Therefore further noise reduction
measures should be implemented to enhance the
ability to stand against noise. In [1], the author
proposed an ACE noise reduction method in
802.11a system. It was deployed by averaging
every two pre-estimated channels in time domain.
In [3], the author applied this method to DVB-T
system. These works are useful to reduce noise.
But both of them can only be applied to time
invariant channel. They appear a performance
floor effect when being used to time-varying
fading channels.
The contribution of this paper is to extend the ACE
algorithm to adapt to time-varying channels and
eliminate the floor effect caused by noise
averaging as in ACE. 1) In order to reflect the
characteristic of the time-varying channel, we
introduce a SNR and speed adaptive factor into
noise reduction algorithm. 2) A novel p-weighted
channel estimation algorithm with noise reduction
is proposed. Different weighting factor means
different tradeoff between noise reduction and
channel tracking. 3) Based on least mean-square
error rule, the optimal value of , which is a
function of pilot structure, Doppler frequency,
noise power and statistical characteristic of time-
varying channel, is deduced. 4) After that, an
approximate expression of is calculated for more
easily application.
The remainder of this paper is organized as follows.
We first introduce the system model and pilot
structure in Section 2. Based on block-type pilot,
p-weighted channel estimation algorithm over
time-varying fading channel is elaborated in
Section 3. Finally, we evaluate the performance of
p-weighted algorithm in Section 4. Section 5
concludes the paper.
2 System description
2.1 Channel model
The wireless fading channel model in time domain
can be expressed as:
(1)
where and are the time-varying amplitude
and propagation delay (assumed that is invariant)
of the ith path, respectively. is the total
number of multipath. The channel correlation
function is given in [2]:
972
___________________________________
978-1-4244-6853-9/10/$26.00 2010 IEEE

(2)
where is the total average power of channel; k,
n denote kth subcarrier and nth symbol
respectively. and are time correlation
function and frequency correlation function.
(3)
(4)
where is the Doppler frequency
normalized by the OFDM symbol rate ,
is the Root-Mean-Square delay-
spread normalized by sampling interval , and
is the zero-order Bessel function of the first
kind.
2.2 Pilot structure
Two types of pilot structures will be involved in
this paper, as shown in Figure 1. The block-type
pilot structure is usually used in systems whose
channel is stable such as WLAN; one of the LTE
pilot structures named interlaced-type pilot
structure usually used in frequency and time fading
channel to obtain frequency and time diversity gain.
(a) block-type pilot (b) one of LTE pilot structures
Figure 1. Two types of pilot structures
3 p-weighted algorithm
As a good designed OFDM system, the ISI can be
eliminated with the cyclic prefix. Therefore the
received signal at the kth subcarrier of the nth
OFDM symbol can be expressed as:
(5)
where is the channel frequency response
(CFR), is the transmitted symbol and
is the zero mean additive white Gaussion
noise with variance .
For simplicity, the block-type pilot structure and
piecewise-linear interpolation are used in the time
domain. Considering two adjacent pilot symbols,
and , the LS estimation can be given as:
(6)
(7)
The variance of is . To reduce noise,
and are weighted by factor
as follows:
(8)
(9)
The weighted noise power is ,
and is related with channel state. When ,
the maximal SNR gain 3dB is obtained, as the
results of [1,3].
After that, we can get the piecewise-linear
interpolation between and (we substitute
with for simplicity).
(10)
In (10), , . Obviously,
when is constant, the estimation value can not
track the state of the time-varying fading channel.
The channel estimation error is:
(11)
The MSE of can be given as (12) [2].
In (12), denotes the expected value of random
variable, denotes the real part of a complex
number, , ,
, .
The overall MSE of the piecewise-linear
interpolation is given in (13).
| |
( )
| | | | ) ( ) 1 (
2
) ( ) 1 ( ) 1 ( 4
) 0 ( )
1
1 )( 1 ( 2 )
1
2 )( 2 2 1 (
3
1
) ( )
1
1 )( 2 2 1 (
2
1
)
1
2 )( 1 (
3
2
) 0 ( )
2
1 (
] ) ( [
1
] [
2
1
0
2 2
2
2 2
2
2
2
2
1
0
2 2
D R p
D
d R
D
d
p
D
d
p
R
D
p p
D
p p
D R
D
p p
D
p p R
D
p
n H E
D
H E
D
d
D
d
9 9
(
+
+
(
+ + + +
9
(
+ + + + + =
A = A
_
_
=
o
To obtain the minimal MSE, a proper value
should be evaluated, as deduced in (14) and (15).
| |
| | { } | |
| | { }
| | { } ) ( ) 0 ( 1
2
3
4
) ( ) 0 ( 1
2
3
2
) (
2
1
4
) ( ) 0 (
2
2
2
2
2
1
0
2
D R R
D
p
D R R
D
d R
D
d
D
D R R
D dp
H dE
D
d
9 + |
.
|
\
|
+ +
9 + |
.
|
\
|
+
9 |
.
|
\
|
9 + =
A
_
=
o
o
(13)
(14)
(12)
973
(15)
-20 -10 0 10 20 30 40
0.5
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
1
Eb/No(dB)
optimal p
approximate p(2)
approximate p(3)
130m/s
70m/s
10m/s
Figure 2. Difference between and approximate
The value expressed by (15) is a function of pilot
structure, Doppler frequency, noise power and
statistical characteristic of time-varying channel.
To make it more practical, we will then deduce an
approximation of the optimal value. Equation
(3) can be developed by series expansion as [4] :
(16)
Generally speaking, the value of is smaller
than 1 in practice because of the good design of
pilot arrangement. For example, is about
0.8311 at speed of 500km/h in LTE system. Hence,
the first two of (16) can be took as an
approximation of .
(17)
Using (15) and (17), we can get the expression of
:
(18)
Figure 2 presents us the difference of and
approximation of at different speed. The blue
line with circle marker is the optimal expressed
by (15), the red line with asterisk maker denotes
the approximation of when the first two of (16)
is taken, as shown in (17); and the pink line with
square maker is the approximation of when the
first three of (16) is taken. We can see that the first
three of (16) is accurate enough to express the
optimal , shown in (18) is more practical for
application, with an acceptable lose in performance
when speed increases.
Figure 3 shows the change of with SNR and
velocity. We can see that, with the increase of SNR
or velocity, moves to 1, otherwise, is closer to
0.5. The reason is that when the SNR or velocity is
high, the accuracy of channel estimation is more
affected by time-variation of channel compared to
channel noise, therefore, moves to 1 for
maintaining the variation of the channel state.
-20
0
20
40
0
50
100
0.5
0.6
0.7
0.8
0.9
1
1.1
SNR(dB)
Velocity(m/s)
p
Figure 3. Relation of , SNR, and velocity
4 Simulations
In this section, we evaluate the performance of
proposed p-weighted algorithm. The simulation
parameters are given in Table 1.Two types of pilot
structures shown in Figure 1(a) and Figure 1(b) are
simulated respectively. The results are given in
Figure 4-11
Table 1. Simulation parameters
parameters values
channel model SCME
multipath 12
max delay 88 samples
bandwidth 20MHz
OFDM subcarriers 2048
virtual carriers 848
sampling frequency 30.72MHz
modulation QPSK
velocity
10m/s
40m/s
carrier frequency 2GHz
In Figure 4-7, the MSE and BER performance of
our p-weighted algorithm, ACE algorithm in [1]
and LS algorithm are compared at the speed of
10m/s and 40m/s (the Doppler frequency is about
66.667Hz and 266.667Hz respectively). The pink
line in Figure 5 and Figure 7 denotes the
performance of ideal channel estimation. It is
obvious that p-weighted and ACE algorithm can
improve SNR when with high noise. But when the
effect of time variation of the channel exceeds the
effect of noise, ACE appears a performance floor
because it eliminates the time variation of channel,
while our p-weighted algorithm compromises
successfully between noise reduction and channel
state tracing.
974
0 5 10 15 20 25 30
10
-4
10
-3
10
-2
10
-1
10
0
fd=66.667Hz(V=10m/s)
Eb/No(dB)
M
S
E
p-weighted
ACE
Linear interpolation
Figure 4. MSE comparison under block-type pilot(10m/s)
0 5 10 15 20 25 30
10
-4
10
-3
10
-2
10
-1
10
0
fd=66.667Hz(V=10m/s)
Eb/No(dB)
B
E
R
p-weighted
ACE
ideal CE
Figure 5. BER comparison under block-type pilot(10m/s)
0 5 10 15 20 25 30
10
-4
10
-3
10
-2
10
-1
10
0
fd=266.667Hz(V=40m/s)
Eb/No(dB)
M
S
E
p-weighted
ACE
Figure 6. MSE comparison under block-type pilot(40m/s)
0 5 10 15 20 25 30
10
-4
10
-3
10
-2
10
-1
10
0
fd=266.667Hz(V=40m/s)
Eb/No(dB)
B
E
R
p-weighted
ACE
ideal CE
Figure 7. BER comparison under block-type pilot(40m/s)
0 5 10 15 20 25 30
10
-4
10
-3
10
-2
10
-1
10
0
fd=66.667Hz(V=10m/s)
Eb/No(dB)
M
S
E
p-weighted
ACE
Figure 8. MSE comparison under interlaced -type
pilot(10m/s)
0 5 10 15 20 25 30
10
-4
10
-3
10
-2
10
-1
10
0
fd=66.667Hz(V=10m/s)
Eb/No(dB)
B
E
R
p-weighted
ACE
ideal CE
Figure 9. BER comparison under interlaced -type
pilot(10m/s)
0 5 10 15 20 25 30
10
-4
10
-3
10
-2
10
-1
10
0
fd=266.667Hz(V=40m/s)
Eb/No(dB)
M
S
E
p-weighted
ACE
Figure 10. MSE comparison under interlaced-type
pilot(40m/s)
0 5 10 15 20 25 30
10
-4
10
-3
10
-2
10
-1
10
0
fd=266.667Hz(V=40m/s)
Eb/No(dB)
B
E
R
p-weighted
ACE
ideal CE
Figure 11. BER comparison under interlaced-type
pilot(40m/s)
975
Figure 8-11 present the system performance when
our p-weighted algorithm is applied to the LTE
pilot structure presented in Figure 1 (b). These
results demonstrate that our algorithm work well
with interlaced-type pilot structure likewise, since
the pilot arrangement design makes the strong
correlation of pilots in frequency domain. In
addition, the interlaced-type pilot performs better
than block-type pilot at low SNR, this is because
the interpolation in frequency domain reduces a
part of noise.
5 Conclusions
In this paper, we proposed a p-weighted noise
reduction algorithm for interpolation of channel
estimation over time-varying channel. Based on
the least mean-square error rule, we deduced the
optimal under block-type pilot and piecewise
linear interpolation in time domain. Both
theoretical analysis and simulation results show p-
weighted algorithm brings a SNR gain at low SNR,
and can trace channel state well at high SNR. For
more easily applying, an approximation of is
deduced, with little lose of performance. Moreover,
though the optimal is obtained under block-type
pilot, we can also apply it to interlaced-type pilot.
The proposed p-weighted algorithm is expected to
improve channel estimation accuracy over both
time-varying channel and time-invariant channel.
Furthermore, the idea of p-weighted algorithm can
be applied to other field where noise reduction of
interpolation is needed.
Acknowledgements
The work presented in this paper was supported in
part by NSFC under Grant 60872049 and
60971082, 863 Program of China under Grant
2007AA10Z235, National Key Basic Research
Program of China(973 Program) 2009CB320407,
National Great Science Specific Project
(2009ZX03003-001 and 2009ZX03003-011), and
Chinese Universities Scientific Fund.
References
[1] W. G. Jeon, K. H. Paik, and Y. S. Cho, An
Efficient Channel Estimation Technique for
OFDM Systems with Transmitter Diversity,
Proc. IEEE Intl. Symp. Personal, Indoor and
Mobile Radio Commun., vol. 2, London, UK,
Sept. 2000, pp. 124650.
[2] C. R. N. Athaudage and A. D. S. Jayalath,
LowComplexity Channel Estimation for
Wireless OFDM Systems, Proc. IEEE Intl.
Symp. Personal, Indoor and Mobile radio
Commun., vol.1, Beijing, China, Sept. 2003.
[3] Y.-S. Lee and H.-N. Kim, Noise Reduction
for Channel Estimation Based on Pilot-Block
Averaging in DVB-T Receivers, IEEE
Transactions on Consumer Electronics,
Vol.52, No. 1, Feb 2006.
[4] M. Abramowitz and I. A. Stegun, Handbook
of Mathematical Functions.Washington, DC:
NBS, 1964.
976

Novel Weigted

Uploaded by

Copyright:

Available Formats

Novel Weigted

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Novel Weigted

Uploaded by

Copyright:

Available Formats

Proceedings of IC-NIDC2010

A NOVEL WEIGHTED CHANNEL ESTIMATION

You might also like