- Email: [email protected]
OFDM frame synchronization based on energy difference of the received preamble
THE JOURNAL OF CHINA UNIVERSITIES OF POSTS AND TELECOMMUNICATIONS Volume 14, Issue 1, March 2007
JIANG Feng, SONG Mei, GUO Da, SONG Jun-de
OFDM frame synchronization based on energy difference
of the received preamble CLC number TN929.05
Article ID 1005-8885 (2007) 01-0096-04
Document A
Abstract The article presents a novel orthogonal frequency
division multiplexing (OFDM) frame synchronization method. The method uses the preamble that contains two identical halves, and the energy difference of the two similar parts of the received preamble in receiver is used to obtain the frame’s correct starting point. Furthennore, an improvement of the proposed method is presented, which uses the preamble that can be divided into four similar portions, and the energy difference of the four similar parts of the received preamble in receiver is used to obtain the frame’s correct starting point. The performances of the frame synchronization methods in multipath channel are compared in t e r n of mean square error (MSE) obtained by simulations. It can be seen from the simulation results that the proposed methods have better performances than Schmidl’s, Minn’s, and Park’s method in multipath channel.
Keywords orthogonal frequency division multiplexing (OFDM) t h i n g offset, frame synchronization
1 lnkaduetlon OFDM is an attractive multicarrier transmission scheme that has high data rate transmission capability and high bandwidth efficiency. So, applications of OFDM to wireless mobile communication systems are investigated by many researchers [I-31. When OFDM technology is applied to wireless mobile communication systems, synchronizations at the receiver is problematic. In general, the received signal is disturbed by multipath fading channel. And the symbol timing offset, the frame timing offset, the sampling clock offset, and carrier frequency offset of a received signal are unknown at the receiver. So synchronization processes such as frame synchronization [4-7 1, symbol synchronization [8, 91, carrier frequency synchronization [ 10-141, and sampling clock synchronization [15] must be executed at the receiver. This
article focuses on frame synchronization because frame synchronization errors would affect all symbols in the frame. Several approaches have been presented to execute frame synchronization [4-71. Among them, Schmidl’s method was proposed initially [41. In the method, a Preamble which contains the same two halves.is used to estimate the frame timing and frequency offset. It is simple and robust to multipath channel. But the timing metric of Schmidl’s method has a plateau, which leads to quite a poorperformance for the frame synchronization. To reduce the uncertainty arising from the plateau of the timing metric, Minn and Park proposed their methods as modifications to Schmidl’s approach [5,6]. They presented two new preambles, which yields sharper timing metrics. While the two methods provide accurate estimates in one-path channel, their performance is bad in the multi-path channel. This article presents a new frame synchronization method, which brings a new sharper timing metric compared with that of Schmidl’s method and exhibits more robustness than Minn’s and Park‘s methods in the multipath channel. Furthermore, an improvement of the proposed method is also presented. The two methods have better performance than Schmidl’s, Minn’s and Park’s method in multipath channel. The OFDM system model and the existing frame synchronization methods are described in Sect. 2. Section 3 covers the proposed method and its improvement. In Sect.4, the advantages of the proposed method and its improvement are demonstrated using the simulation results. Finally, the conclusion is given in Sect. 5.
2 OFDM tlmlng synchronlzatlon 2.1 OFDM system model It is assumed that the nth received sample is h(m)x(n- rn - z,,)
y ( n )=
(1)
m=0
Received date: 2006-05-08 JIANG Feng (9 ), SONG Mei, GUO Da, SONG Jun-de CAD and PCN Center, Electronic Engineering Institute,Beijing University of Posts and Telecommunications. Beijing 100876, China E-mail: [email protected]
where Z, is the delay of the path m , h ( x ) is the channel impulse response, whose memory is L , and x(n) is the time-domain transmitted OFDM signal. At the receiver, the received signal with frequency offset can
No. 1
JIANG Feng, et al.: OFDM frame synchronization based on energy difference of the received preamble
be expressed as
r(n) = y(n)eX2xfmn++) + w(n) (2) where f, is the frequency offset, @ is the channel initial phase, and w(n) is the Gaussian white noise.
97
modified preamble[CN,, DN14C i I 2 D J I j 2 ] ,where DNI4 is designed to be symmetric with C,,, . Then the timing metric is
2.2 OFDM frame synchronlzatlon
and The goal of OFDM frame synchronization is to obtain the frame’s correct starting point. And before executing other tasks, such as frequency synchronization, channel estimation, etc., which is necessary to OFDM system, the frame’s correct starting point must be obtained. Several approaches have been presented to execute frame synchronization in Refs. [4-6]. 1) Schmidl’s method in Ref. [4],Schmidl used the time domain preamble that contains the same two halves to obtain frame timing offset: [ A N t 2AN/*], where N represents the preamble that contains N samples and ANI2 represents samples of length N / 2 . Schmidl searched the maximum point of the timing metric defined as Eq. (3) to obtain the starting point of the symbol. (3) where NI2-I
(4) k=O
The frame’s correct starting point nEcan be obtained as =argmdax(&(dN
(12)
k=O N12
R,(d)=CIr(d+k)12
(13)
k =O
The timing offset n, can be estimated as
(M,(d))
%
(14)
Both the Minn’s and Park’s methods have their peaks at the correct starting point in one-path channel since they have used special preambles. Minn’s method used two negative-valued repeated segments in its preamble, and Park‘s method used the four inverse repeated segments. Correlation of these negative or inverse repeated segments results in a decrease of the timing metric at an incorrect OFDM symbol starting point. However, in spite of the reduction of the timing metric plateau, the inequality of these negative or inverse repeated segments results in a poor performance in the multipath channel. It is observed that the MSE of Minn’s estimation is quite large in multipath channels, as can be seen from the results in Ref. [5]
8 Proposedmethods
k=O
nc
N12
P,(d)=Zr(d-k)r(d+k)
(6)
Obviously, with the help of cyclic prefix and two A,,, , Schmidl’s method is robust to multipath channel [4].But the timing metric of Schmidl’s method has its peak for the entire interval of the cyclic prefix, which leads to some uncertainty regarding the starting point of the OFDM symbol. Thus Minn and Park presented their improved methods. 2) Minn’s method: in Ref. [5],Minn proposed a modified preamble[B,,, BN,4-BN,4 -BN14].Then the timing metric is (7)
where
The reason for Minn’s and Park’s method having better performances than Schmidl’s in one path channel is that they have sharper peaks than Schmidl’s. And the reason for their poor performance to multipath channel is that their preambles are not similar to Schmidl’s preamble, which can be split into two or more identical portions. For these two reasons, we initially propose a new frame synchronization method using the same preamble with Schmidl’s method. And the method has a sharper peak than Schmidl’s even in the channel with ISI. Second, an improved method using a new preamble is presented. In this article, the two methods are named the basic method and the improved method. 1) The basic method: First, we use the preamble [E,,, E ~ , ~The ] .new time timing metric is
(9)
where The frame’s correct starting point n, can be obtained as nE = arg
7( M m
NIZ-l
(10)
3) Park’s Method: in Ref. [6], Park also proposed another
D(d)= C k =O
e ( d ) is defied as Eq. (4).
)-2Ie(d)l
(16)
2007
The Journal of CHUPT
98
The frame’s correct starting point n, can be obtained as
n, = arg=
(17)
(ML(d))
Without noise, the timing metric Eq. (15) has a domain, in which D,(d)=O and out of which ID,(d)l>O , in the interval of the cyclic prefix. So at the end point of the domain, which is the c o m t starting point, M L + m , That is to say, the proposed timing metric achieves its peak at the correct starting point. Thus, the timing metric plateau is reduced. Figures 1 and 2 show two examples of the timing metrics. Figure 1 is under no noise and one-path channel distortion with 1 024 subcarriers and 128 cyclic prefix. The correct timing point is indexed as 0 in the figure. And the proposed timing metric Eq.(15) is compared to those of Schmidl’s, Minn’s and Parks. Figure 2 is under the channel with two paths that have the same path gains and path delays 0,80 samples. And the other parameters in Fig. 2 are similar to that of Fig. 1.
creates a plateau in both channels and in the channel with one path; Minn’s and Park’s methods reduce the plateau and yield sharp timing metrics. But in the channel with two paths, Minn’s method also has a timing metric plateau and Park’s method’s peak is not at the correct timing point, which means Minn’s and Park‘s methods probably have worse performances in the multipath channel than in the one-path channel. And the basic method has one sharp peak in each channel, which means that the basic method is robust to IS1 and is only influenced by noise. But the results of the basic method are interfered by the noise of the entire interval of the preamble. So an improvement, the results of which are influenced by the noise of the partial interval of the preamble, is presented in the following. 2) The improved method: We propose an improvement of the basic method using the preamble [F,,, F,,, FN14F N I 4 ] . And the improved time timing metric is
- - Schmidl method where
I -300
-200
-100
‘*-%-
0
100
zoo
300
Timeisample
N14-I
p,‘(d)=
-The proposed
r*(d+k)r k =O
.-
k=O
t- -300
-100
-200
0
100
200
300
Time /sample
The frame’s correct starting point nE can be obtained as
n, = arg y ( M & ( d ) )
Fig. 1 Comparison of the timing metric in the channel with one path
(23)
Theoretically, the peaks of those Mrare at the same point,
1.0
which is the correct starting point. But the noise would make those peaks not always at the correct starting point. And M,‘ is only influenced by the noise of the interval of the i th and
’2G 0.5
the last FN,4( i = 1,2,3 ). If the noise of the interval of certain FNI4 results in erratic corresponding M’ values and the other
eE
- Schmial method - - - - Minn method
M
0
-300
-200
0
-100
100
200
300
values are not error-prone, the final M& maybe free of errors, by using the arithmetic geometric mean of all MI.In contrast,
Timeisample
the noise of the interval of certain FN,4 maybe directly results in erratic final M i m values in the basic method. So the
The proposed metbod
performance of the improved method is better than the basic method.
E
-300
-200
-100
0
100
200
300
4 Qmulatlonresutb
Time isample
Fig. 2
Comparison of the timing metric in the channel with two path
From Figs. 1 and 2, it can be seen that Schmidl’s method
In this section, numerical results are presented to illustrate the advantages of the two proposed methods in terms of MSE. The simulation results are averaged over 1 OOO Monte Car10 runs.
No. 1
JIANG Feng, et al.: OFDM frame synchronization based on energy difference of the received preamble
And the OFDM system with 64 subcarriers and 16 cyclic prefix is considered. The multipath channel is modeled as 6 , paths with path delays r,,,,0 , 3 , 6 , 9 , 12, 15 samples and path gains given by
“=/p e-z” 115
(24)
mE [I, 61
Figure 3 shows the MSE of the methods in multipath channel. Figure 3 shows that the two proposed methods have better performance than Schmidl’s, Minn’s and Park‘s method in multipath channel. It is also shown in Fig. 3 that the improved method has better performance than the basic method. h
2 102
P
lo’
M
;E
100
* c
8
4. Schmidl T M, Cox D C. Robust frequency and timing synchronization for OFDM. E E E Trans. Commun, 1997, 45: 1613-162 1 5. Minn H,Zeng M, Bhargava V K. On timing offset estimation for OFDM systems. IEEECommun. Lett,2000,4: 242-244 6. Park B, Cheon H, Kang C, et al.. A novel timing estimation method for OFDM systems. IEEE Comm. Lett, 2003, 7(5): 239-241 7. Wang Zhi-xin, Li Jim-dong, Chen Chen, at al. An efficient symbol timing and frequency synchronization scheme for OFDM systems. Journal of Chongqing University of Posts and Telecommunications,2005, 17(2): 147-152 (in Chinese) 8. Zhou En, Chen Mao-mao, Wang Wen-bo. A novel timing estimation method for OFDM systems over multipath fading channels. Journal of Beijing University of Posts and Telecommunications,2005,28 (3): 62-64 (in Chinese) 9. Takahashi K, Saba T. A novel symbol synchronization algorithm with reduced influence of IS1 for OFDM systems. IEEE GLOBECOM, 2001: 524-528
Y
#
99
1
10-1
+- Schmidl method -B- Park method +Minn method -A-
The basic method
-e- The improved method
~
5
10
15
20
25
30
EdNO
Fig. 3 MSE of estimators in multipath channel
5 Condudons Two frame synchronization methods are presented in this artide. And the proposed methods sharpen the timing metTic even in multipath channel, and have better performances than Schmidl’s ,Minn’s and Park‘s methods. Acknowledgements This work is supported by the National Natural Seience Foundation of China (60572119).
References Richard V Ne, Ramjee P. OFDM for wireless multimedia communications. Artech House, 2000: 25-32 Heidi S , Marc M. Analysis and optimization of the performance of OFDM on frequency-selective time-selective fading channel. IEEETrans. C o m u n , 1999,47(12): 1811-1819 Shinsuke H, Ramjee P. Overview of multicarrier CDMA. IEEE Commun. Mag, 1997,25 (3): 126-133
10. Moose P H. A technique for orthogonal frequency division multiplexing frequency offset correction. IEEE Trans. Commun, 1994,42: 2908-2914 11. Luo Isem. Carrier frequency acquisitionand tracking for OFDM systems. IEEE Trans. Comm, 1996,44: 1590-1598 12. Van de Beek I 1. ML estimation of time and frequency offset in OFDM systems. IEEE Transactions on Signal Processing, 1997, 45(7): 1800-1805 13. Tureli U, Liu H, Zoltowski M D. OFDM blind carrier offset estimation: ESPRIT. E E E Transactions on Communications, 2000,48(9): 1459-1461 14. Wang Chun-guang, Zhou Zheng. A carrier synchronization algorithm for IEEE 802.11a system. The Journal of China Universities of Posts and Telecommunicans,2004.1 l(S1): 62-66 15. Michael S , Stefan A. F, GUMX F, et al. Optimum receiver design for wireless broad-band systems using OFDM part I. IEEE Transactions on Communications,1999,47(11): 1668-1677
Biography: JIANG Feng, Ph. D. Candidate, Beijing University of Posts and Telecommunications, research dompin include future broadband telecom, wireless telecom, such as OFDM and MIMO, etc.
JIANG Feng, SONG Mei, GUO Da, SONG Jun-de
OFDM frame synchronization based on energy difference
of the received preamble CLC number TN929.05
Article ID 1005-8885 (2007) 01-0096-04
Document A
Abstract The article presents a novel orthogonal frequency
division multiplexing (OFDM) frame synchronization method. The method uses the preamble that contains two identical halves, and the energy difference of the two similar parts of the received preamble in receiver is used to obtain the frame’s correct starting point. Furthennore, an improvement of the proposed method is presented, which uses the preamble that can be divided into four similar portions, and the energy difference of the four similar parts of the received preamble in receiver is used to obtain the frame’s correct starting point. The performances of the frame synchronization methods in multipath channel are compared in t e r n of mean square error (MSE) obtained by simulations. It can be seen from the simulation results that the proposed methods have better performances than Schmidl’s, Minn’s, and Park’s method in multipath channel.
Keywords orthogonal frequency division multiplexing (OFDM) t h i n g offset, frame synchronization
1 lnkaduetlon OFDM is an attractive multicarrier transmission scheme that has high data rate transmission capability and high bandwidth efficiency. So, applications of OFDM to wireless mobile communication systems are investigated by many researchers [I-31. When OFDM technology is applied to wireless mobile communication systems, synchronizations at the receiver is problematic. In general, the received signal is disturbed by multipath fading channel. And the symbol timing offset, the frame timing offset, the sampling clock offset, and carrier frequency offset of a received signal are unknown at the receiver. So synchronization processes such as frame synchronization [4-7 1, symbol synchronization [8, 91, carrier frequency synchronization [ 10-141, and sampling clock synchronization [15] must be executed at the receiver. This
article focuses on frame synchronization because frame synchronization errors would affect all symbols in the frame. Several approaches have been presented to execute frame synchronization [4-71. Among them, Schmidl’s method was proposed initially [41. In the method, a Preamble which contains the same two halves.is used to estimate the frame timing and frequency offset. It is simple and robust to multipath channel. But the timing metric of Schmidl’s method has a plateau, which leads to quite a poorperformance for the frame synchronization. To reduce the uncertainty arising from the plateau of the timing metric, Minn and Park proposed their methods as modifications to Schmidl’s approach [5,6]. They presented two new preambles, which yields sharper timing metrics. While the two methods provide accurate estimates in one-path channel, their performance is bad in the multi-path channel. This article presents a new frame synchronization method, which brings a new sharper timing metric compared with that of Schmidl’s method and exhibits more robustness than Minn’s and Park‘s methods in the multipath channel. Furthermore, an improvement of the proposed method is also presented. The two methods have better performance than Schmidl’s, Minn’s and Park’s method in multipath channel. The OFDM system model and the existing frame synchronization methods are described in Sect. 2. Section 3 covers the proposed method and its improvement. In Sect.4, the advantages of the proposed method and its improvement are demonstrated using the simulation results. Finally, the conclusion is given in Sect. 5.
2 OFDM tlmlng synchronlzatlon 2.1 OFDM system model It is assumed that the nth received sample is h(m)x(n- rn - z,,)
y ( n )=
(1)
m=0
Received date: 2006-05-08 JIANG Feng (9 ), SONG Mei, GUO Da, SONG Jun-de CAD and PCN Center, Electronic Engineering Institute,Beijing University of Posts and Telecommunications. Beijing 100876, China E-mail: [email protected]
where Z, is the delay of the path m , h ( x ) is the channel impulse response, whose memory is L , and x(n) is the time-domain transmitted OFDM signal. At the receiver, the received signal with frequency offset can
No. 1
JIANG Feng, et al.: OFDM frame synchronization based on energy difference of the received preamble
be expressed as
r(n) = y(n)eX2xfmn++) + w(n) (2) where f, is the frequency offset, @ is the channel initial phase, and w(n) is the Gaussian white noise.
97
modified preamble[CN,, DN14C i I 2 D J I j 2 ] ,where DNI4 is designed to be symmetric with C,,, . Then the timing metric is
2.2 OFDM frame synchronlzatlon
and The goal of OFDM frame synchronization is to obtain the frame’s correct starting point. And before executing other tasks, such as frequency synchronization, channel estimation, etc., which is necessary to OFDM system, the frame’s correct starting point must be obtained. Several approaches have been presented to execute frame synchronization in Refs. [4-6]. 1) Schmidl’s method in Ref. [4],Schmidl used the time domain preamble that contains the same two halves to obtain frame timing offset: [ A N t 2AN/*], where N represents the preamble that contains N samples and ANI2 represents samples of length N / 2 . Schmidl searched the maximum point of the timing metric defined as Eq. (3) to obtain the starting point of the symbol. (3) where NI2-I
(4) k=O
The frame’s correct starting point nEcan be obtained as =argmdax(&(dN
(12)
k=O N12
R,(d)=CIr(d+k)12
(13)
k =O
The timing offset n, can be estimated as
(M,(d))
%
(14)
Both the Minn’s and Park’s methods have their peaks at the correct starting point in one-path channel since they have used special preambles. Minn’s method used two negative-valued repeated segments in its preamble, and Park‘s method used the four inverse repeated segments. Correlation of these negative or inverse repeated segments results in a decrease of the timing metric at an incorrect OFDM symbol starting point. However, in spite of the reduction of the timing metric plateau, the inequality of these negative or inverse repeated segments results in a poor performance in the multipath channel. It is observed that the MSE of Minn’s estimation is quite large in multipath channels, as can be seen from the results in Ref. [5]
8 Proposedmethods
k=O
nc
N12
P,(d)=Zr(d-k)r(d+k)
(6)
Obviously, with the help of cyclic prefix and two A,,, , Schmidl’s method is robust to multipath channel [4].But the timing metric of Schmidl’s method has its peak for the entire interval of the cyclic prefix, which leads to some uncertainty regarding the starting point of the OFDM symbol. Thus Minn and Park presented their improved methods. 2) Minn’s method: in Ref. [5],Minn proposed a modified preamble[B,,, BN,4-BN,4 -BN14].Then the timing metric is (7)
where
The reason for Minn’s and Park’s method having better performances than Schmidl’s in one path channel is that they have sharper peaks than Schmidl’s. And the reason for their poor performance to multipath channel is that their preambles are not similar to Schmidl’s preamble, which can be split into two or more identical portions. For these two reasons, we initially propose a new frame synchronization method using the same preamble with Schmidl’s method. And the method has a sharper peak than Schmidl’s even in the channel with ISI. Second, an improved method using a new preamble is presented. In this article, the two methods are named the basic method and the improved method. 1) The basic method: First, we use the preamble [E,,, E ~ , ~The ] .new time timing metric is
(9)
where The frame’s correct starting point n, can be obtained as nE = arg
7( M m
NIZ-l
(10)
3) Park’s Method: in Ref. [6], Park also proposed another
D(d)= C k =O
e ( d ) is defied as Eq. (4).
)-2Ie(d)l
(16)
2007
The Journal of CHUPT
98
The frame’s correct starting point n, can be obtained as
n, = arg=
(17)
(ML(d))
Without noise, the timing metric Eq. (15) has a domain, in which D,(d)=O and out of which ID,(d)l>O , in the interval of the cyclic prefix. So at the end point of the domain, which is the c o m t starting point, M L + m , That is to say, the proposed timing metric achieves its peak at the correct starting point. Thus, the timing metric plateau is reduced. Figures 1 and 2 show two examples of the timing metrics. Figure 1 is under no noise and one-path channel distortion with 1 024 subcarriers and 128 cyclic prefix. The correct timing point is indexed as 0 in the figure. And the proposed timing metric Eq.(15) is compared to those of Schmidl’s, Minn’s and Parks. Figure 2 is under the channel with two paths that have the same path gains and path delays 0,80 samples. And the other parameters in Fig. 2 are similar to that of Fig. 1.
creates a plateau in both channels and in the channel with one path; Minn’s and Park’s methods reduce the plateau and yield sharp timing metrics. But in the channel with two paths, Minn’s method also has a timing metric plateau and Park’s method’s peak is not at the correct timing point, which means Minn’s and Park‘s methods probably have worse performances in the multipath channel than in the one-path channel. And the basic method has one sharp peak in each channel, which means that the basic method is robust to IS1 and is only influenced by noise. But the results of the basic method are interfered by the noise of the entire interval of the preamble. So an improvement, the results of which are influenced by the noise of the partial interval of the preamble, is presented in the following. 2) The improved method: We propose an improvement of the basic method using the preamble [F,,, F,,, FN14F N I 4 ] . And the improved time timing metric is
- - Schmidl method where
I -300
-200
-100
‘*-%-
0
100
zoo
300
Timeisample
N14-I
p,‘(d)=
-The proposed
r*(d+k)r k =O
.-
k=O
t- -300
-100
-200
0
100
200
300
Time /sample
The frame’s correct starting point nE can be obtained as
n, = arg y ( M & ( d ) )
Fig. 1 Comparison of the timing metric in the channel with one path
(23)
Theoretically, the peaks of those Mrare at the same point,
1.0
which is the correct starting point. But the noise would make those peaks not always at the correct starting point. And M,‘ is only influenced by the noise of the interval of the i th and
’2G 0.5
the last FN,4( i = 1,2,3 ). If the noise of the interval of certain FNI4 results in erratic corresponding M’ values and the other
eE
- Schmial method - - - - Minn method
M
0
-300
-200
0
-100
100
200
300
values are not error-prone, the final M& maybe free of errors, by using the arithmetic geometric mean of all MI.In contrast,
Timeisample
the noise of the interval of certain FN,4 maybe directly results in erratic final M i m values in the basic method. So the
The proposed metbod
performance of the improved method is better than the basic method.
E
-300
-200
-100
0
100
200
300
4 Qmulatlonresutb
Time isample
Fig. 2
Comparison of the timing metric in the channel with two path
From Figs. 1 and 2, it can be seen that Schmidl’s method
In this section, numerical results are presented to illustrate the advantages of the two proposed methods in terms of MSE. The simulation results are averaged over 1 OOO Monte Car10 runs.
No. 1
JIANG Feng, et al.: OFDM frame synchronization based on energy difference of the received preamble
And the OFDM system with 64 subcarriers and 16 cyclic prefix is considered. The multipath channel is modeled as 6 , paths with path delays r,,,,0 , 3 , 6 , 9 , 12, 15 samples and path gains given by
“=/p e-z” 115
(24)
mE [I, 61
Figure 3 shows the MSE of the methods in multipath channel. Figure 3 shows that the two proposed methods have better performance than Schmidl’s, Minn’s and Park‘s method in multipath channel. It is also shown in Fig. 3 that the improved method has better performance than the basic method. h
2 102
P
lo’
M
;E
100
* c
8
4. Schmidl T M, Cox D C. Robust frequency and timing synchronization for OFDM. E E E Trans. Commun, 1997, 45: 1613-162 1 5. Minn H,Zeng M, Bhargava V K. On timing offset estimation for OFDM systems. IEEECommun. Lett,2000,4: 242-244 6. Park B, Cheon H, Kang C, et al.. A novel timing estimation method for OFDM systems. IEEE Comm. Lett, 2003, 7(5): 239-241 7. Wang Zhi-xin, Li Jim-dong, Chen Chen, at al. An efficient symbol timing and frequency synchronization scheme for OFDM systems. Journal of Chongqing University of Posts and Telecommunications,2005, 17(2): 147-152 (in Chinese) 8. Zhou En, Chen Mao-mao, Wang Wen-bo. A novel timing estimation method for OFDM systems over multipath fading channels. Journal of Beijing University of Posts and Telecommunications,2005,28 (3): 62-64 (in Chinese) 9. Takahashi K, Saba T. A novel symbol synchronization algorithm with reduced influence of IS1 for OFDM systems. IEEE GLOBECOM, 2001: 524-528
Y
#
99
1
10-1
+- Schmidl method -B- Park method +Minn method -A-
The basic method
-e- The improved method
~
5
10
15
20
25
30
EdNO
Fig. 3 MSE of estimators in multipath channel
5 Condudons Two frame synchronization methods are presented in this artide. And the proposed methods sharpen the timing metTic even in multipath channel, and have better performances than Schmidl’s ,Minn’s and Park‘s methods. Acknowledgements This work is supported by the National Natural Seience Foundation of China (60572119).
References Richard V Ne, Ramjee P. OFDM for wireless multimedia communications. Artech House, 2000: 25-32 Heidi S , Marc M. Analysis and optimization of the performance of OFDM on frequency-selective time-selective fading channel. IEEETrans. C o m u n , 1999,47(12): 1811-1819 Shinsuke H, Ramjee P. Overview of multicarrier CDMA. IEEE Commun. Mag, 1997,25 (3): 126-133
10. Moose P H. A technique for orthogonal frequency division multiplexing frequency offset correction. IEEE Trans. Commun, 1994,42: 2908-2914 11. Luo Isem. Carrier frequency acquisitionand tracking for OFDM systems. IEEE Trans. Comm, 1996,44: 1590-1598 12. Van de Beek I 1. ML estimation of time and frequency offset in OFDM systems. IEEE Transactions on Signal Processing, 1997, 45(7): 1800-1805 13. Tureli U, Liu H, Zoltowski M D. OFDM blind carrier offset estimation: ESPRIT. E E E Transactions on Communications, 2000,48(9): 1459-1461 14. Wang Chun-guang, Zhou Zheng. A carrier synchronization algorithm for IEEE 802.11a system. The Journal of China Universities of Posts and Telecommunicans,2004.1 l(S1): 62-66 15. Michael S , Stefan A. F, GUMX F, et al. Optimum receiver design for wireless broad-band systems using OFDM part I. IEEE Transactions on Communications,1999,47(11): 1668-1677
Biography: JIANG Feng, Ph. D. Candidate, Beijing University of Posts and Telecommunications, research dompin include future broadband telecom, wireless telecom, such as OFDM and MIMO, etc.