應用於 OFDM 系統之強健化 內部接收機架構設計

Post on 15-Jan-2016

208 views 1 download

description

應用於 OFDM 系統之強健化 內部接收機架構設計. 指導老師:高永安 學 生:蘇家弘. A Robust Inner Receiver Structure Design for OFDM Systems. Outline. OFDM system block diagram OFDM baseband signal model Inner receiver structure Channel estimation LMS algorithm Selection of  Pilot-based phase estimator - PowerPoint PPT Presentation

Transcript of 應用於 OFDM 系統之強健化 內部接收機架構設計

1

應用於 OFDM 系統之強健化內部接收機架構設計

指導老師 :高永安學 生 :蘇家弘

A Robust Inner Receiver Structure Design for OFDM Systems

2

Outline OFDM system block diagram OFDM baseband signal model Inner receiver structure

Channel estimation LMS algorithm Selection of Pilot-based phase estimator

Dynamic simulation by Simulink 5.0 Conclusion and future work

3

Serial to

Parallel

Parallelto

Serial IFFT

D/AConver

ter

CH

Parallelto

Serial

Serialto

ParallelFFT

A/DConver

ter

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

TxData

Transmitter

Receiver

RxData yn,l

xn,l

Yk,l

Xk,l

OFDM system block diagram

EqEq...

Eq

SFO

Up convert

Down convert

CFO

n: n-th sample pointk: k-th subcarrierl: l-th subcarrier

4

Carrier Frequency Offsets CFO is due to the

oscillator mismatch from up convert and down convert

f

CFO simulation

5

CFO calculation for IEEE 802.11a Maximum quantity of CFO = 20ppm for

5GHz

k: k-th subcarrier, l: l-th OFDM symbol , N=64, n=80

620 20 10 250 1000.32

312.5 312.5

ppm

kHz

2 n l

N

6

Sampling Frequency Offsets SFO is caused by the

oscillator mismatch between A/D & D/A converter

SFO simulation

t

TTX

TRX

When TRX > TTX

7

SFO calculation for IEEE 802.11a

TTX=1/(20MHz 400Hz),

TRX=1/(20MHz 400Hz) k: k-th subcarrier, l: l-th OFDM symbol ,

N=64, n=80

2 ( )180TX RX

SFOTX

k T T l

NT

8

OFDM baseband signal model OFDM baseband signal after IFFT at the

transmitter side The received OFDM baseband signal

before FFT -------- (2)

12 /

, ,0

1 Nj kn N

n l k lk

x X eN

12 /

, ,0

1 Nj kn N

n l k lk

y Y eN

n: n-th sample pointk: k-th subcarrierl: l-th subcarrier

-------- (1)

9

The received OFDM signal is influenced by channel effect, residual CFO, SFO, initial symbol timing offset and before FFT we can describe (2) as follows:

-------- (3)

Td : initial symbol timing offset

Hk : frequency response of channel : residual CFO : initial phase offset Ts : sampling clock period at the transmitter

Ts’: sampling clock period at the receiver

''( )( )

{2 [ ( ) ] }

, ,

, ,

d s s ss s k

u u

kT k T T N l Gj f N l G T

T Tk l k l k

k l k l

Y X H e

N I

OFDM baseband signal model

f

k

CFOSFO

10

OFDM baseband signal model The ICI produced by residual CFO is much

smaller compared to Gaussian noise. N’k,l combine Ik,l and Nk,l

and (3) can be modified as:

-------- (4)

''( )( )

{2 [ ( ) ] }

, ,

arg[ ],

d s s ss s k

u u

k

kT k T T N l Gj f N l G T

T Tk l k l k

j Hk l

Y X e H

e N

,k lN

11

OFDM baseband signal model The effect of CFO and SFO can be

represented as :

-------- (5)

and

'' '

,

'

( )( )2 [ ( ) ]s s

k l s ku

k k

k T T N G lf N G lT

T

l

' arg[ ] 2 /k k k d uH kT T

12

The difference between inner and outer receiver M. Speth, S. A. Fechtel, G. Fock and H. Meyr, “Optimum Receiver Design for

Wireless Broad-band Systems Using OFDM-Part II,” IEEE Trans. Commun., vol. 49, pp.571-578, Apr. 2001.

Decoding &demodulation

13

Inner receiver structure

Frame Detection

Carrier FrequencyOffset Estimation

Symbol Timing

Buffer

Frequency OffsetCompensation

Remove Prefix

S/P

Input signal

FFT

14

Inner receiver structure

,k ld,k lY

FFTInitial coefficient

FrequencyDomain

Equalizer

Pilot-based phase estimator

Phase compensation

Outer receiver

Hard decision

Update coefficientof equalizer

Phase compensation

Training sequenceData

Pilot

,k lX

15

Channel estimation by least square error Lk,l : transmitted training sequence

Rk,l : received training sequence

: equalized training sequenceHk : channel

Nk,l : noise

: equalizer initial coefficient Equalized training sequence

,k lR

,eq kH

, , ,

, , ,

k l k l eq k

k l k k l eq k

R R H

L H N H

In 802.11al : 2 long training symbol

k : 52 subcarrier

16

Channel estimation by least square error Error between transmitted signal and

equalized signal

Find optimal Heq,k for minimum value of ek

Setting the partial derivative of ek

2 2

,1 ,1 ,2 ,2k k k k ke R L R L

,1 ,1 ,2 ,2

2 2

,1 ,2

k k k keq

k k

L R L RH

R R

17

LMS algorithm Filtering output: Yk=wk

HXk

Error estimation: ek=dk-Yk Tap-weight vector adaptation

*, 1 , , ,k l k l k k l k lw w e X

*

k is the step size that modified by channel condition

After hard decision

18

selection of Normalized-LMS & time average

, 1 , , ,

, ,

, 1 , 1

1

0

1 1

k l k l k k l k l

kk

k l k l

kk k

k l k l

w w l e Y

R R

l lY Y

0 < < 1

Training sequence

19

Pilot-based phase estimator

After giving the

appropriate weight

Re

Im

Re

Im

Received pilots

∠ 1 ∠ 2

Maximum ratio combination (MRC) pilot

O

A

B

A’

B’

O

C

C’

20

Simulation by Simulink 5.0

21

22

Unequalized signal spacing plot

Channel A (Ts=50ns, TRMS=50ns )SNR=10dBResidual CFO =0.01 SFO=800Hz (Ts=1/(20MHz-400), =1/(20MHz+400 )Code rate=1/2, QPSK44 OFDM symbol per packet1000 packet

23

After channel equalization

Applied the proposed inner receiver structure

24

IEEE 802.11a PER v.s. SNR =0.15=0.3Channel A (Ts=50ns, TRMS=50ns )Residual CFO =0.01 SFO=800Hz (Ts=1/(20MHz-400), =1/(20MHz+400 )PSDU=256 bytes1000 packet

sT

25

IEEE 802.11a PER v.s. SNR Channel B (Ts=50ns, TRMS=100ns ) Channel C (Ts=50ns, TRMS=150ns )

26

IEEE 802.11a PER v.s. SNR Channel D (Ts=50ns, TRMS=200ns ) Channel E (Ts=50ns,

TRMS=250ns )

27

PER v.s. SNR with different

=0.3Channel A (Ts=50ns, TRMS=50ns )Residual CFO =0.01 SFO=800Hz (Ts=1/(20MHz-400), =1/(20MHz+400 )44 OFDM symbol per packet1000 packet

sT

28

PER v.s. SNR with different

Channel C (Ts=50ns, TRMS=150ns ) Channel E (Ts=50ns, TRMS=250ns )

29

PER v.s. SNR with different

=0.3Channel A (Ts=50ns, TRMS=50ns )Residual CFO =0.01 SFO=800Hz (Ts=1/(20MHz-400), =1/(20MHz+400 )200 OFDM symbol per packet1000 packet

sT

30

PER v.s. SNR with different

Channel C (Ts=50ns, TRMS=150ns ) Channel E (Ts=50ns, TRMS=250ns )

31

PER v.s. with different channel

=0.3Channel A (Ts=50ns, TRMS=50ns )Channel C (Ts=50ns, TRMS=150ns )Channel E (Ts=50ns, TRMS=250ns )SNR=10dBResidual CFO =0.01 SFO=800Hz (Ts=1/(20MHz-400), =1/(20MHz+400 )44 OFDM symbol per packet1000 packet

sT

32

PER v.s. with different channel

=0.3Channel A (Ts=50ns, TRMS=50ns )Channel C (Ts=50ns, TRMS=150ns )Channel E (Ts=50ns, TRMS=250ns )SNR=10dBResidual CFO =0.01 SFO=800Hz (Ts=1/(20MHz-400), =1/(20MHz+400 )200 OFDM symbol per packet1000 packet

sT

33

A Robust Inner Receiver Structure

Simulink 5.0

Conclusion

pilot-based phase estimator

2. Assist the LMS equalizer in phase tracking

1. Compensate the residual CFO

Dynamic simulation

34

Future work At present

Frequency selective fading channel LMS algorithm

The future work Slow fading channel Other adaptive algorithms Decoding block of Simulink 5.0

35

Reference IEEE Std 802.11a-1999, Part 11: Wireless LAN Medium Access Control (MAC) and

Physical Layer (PHY) specifications: High Speed Physical Layer in the 5GHz Band. Yung-An Kao; Chia-Hung Su; Shih-Kai Lee; Chung-Lung Hsiao; Po-Lin Chio, “A

robust design of inner receiver structure for OFDM systems,” IEEE Conference. on Consumer Electronics, pp. 377-378, Jan. 2005.

S. Haykin, Adaptive Filter Theory, Englewood Cliffs, NJ: Prentice-Hall, 2002, 4 th Ed. M. Speth, S. A. Fechtel, G. Fock and H. Meyr, “Optimum Receiver Design for Wireless

Broad-band Systems Using OFDM-Part I,” IEEE Trans. Commun., vol. 47, pp. 1668-1677, Nov. 1999.

M. Speth, S. A. Fechtel, G. Fock and H. Meyr, “Optimum Receiver Design for Wireless Broad-band Systems Using OFDM-Part II,” IEEE Trans. Commun., vol. 49, pp.571-578, Apr. 2001.

Doufexi, A.; Armour, S.; Butler, M.; Nix, A.; Bull, D.; McGeehan, J.; Karlsson, P., “A comparison of the HIPERLAN/2 and IEEE 802.11a wireless LAN standards,” IEEE Magazine on Comm. Vol. 40, pp.172-180, May 2002.

黃凡維 , 2004, “ 一揭最小均方差頻域等化器應用於正交分頻多工系統之特性分析 ,” 長庚大學電機工程研究所碩士論文

36

Any Questions?

37

Channel estimation by LSE

2 2

,1 ,1 ,2 ,2

*,1 ,1 ,1 ,1*

,1 ,1 ,1 ,1

*,2 ,2 ,2 ,2*

,2 ,2 ,2 ,2

( ) ( )( ) ( )

( ) ( )( ) ( )

k k k k

eq eq

k k k kk k k k

eq eq

k k k kk k k k

eq eq

R L R L

H H

R L R LR L R L

H H

R L R LR L R L

H H

2 2

,1 ,1 ,2 ,2k k k k ke R L R L

38

Channel estimation by LSE Applying we obtain

and

eq

CH

eqR IeqH H

, ,,

, ,

1 ( ) 1 1 0eq

eq k eq kCH eq k

eq k eq kR I

H HH j j j

H H

* *, ,*

,, ,

1 ( ) 1 1 2eq

eq k eq kCH eq k

eq k eq kR I

H HH j j j

H H

R:real partI:imaginary part

39

Channel estimation by LSE* *

,1 ,1 ,2 ,2,1 ,1 ,2 ,2

* *1, , 1, 2, , 2,

1, , 1, 2, , 2,

1, , 1, 1, 2, ,

( ) ( )( ) 0 ( ) 0

( ) ( )( ) ( )

( ) (

k k k kk k k k

eq eq

k eq k k k eq k kk eq k k k eq k k

eq eq

k eq k k k k eq k

R L R LR L R L

H H

R H L R H LR H L R H L

H H

R H L R R H

2, 2,)k kL R

,1 ,1 ,2 ,2

2 2

,1 ,2

k k k keq

k k

L R L RH

R R

40

Comparison between the MRC pilot and the original pilot multiplication

MRC pilot 21 7 7 21

, , ,CFO SFO CFO SFO CFO SFO CFO SFOj j j jA B Ce De e e

21 7 7 21CFO SFO SFO SFO SFO CFO SFOj j j j j j je e e eA B C EeDe e

21 7 7 21CFO SFO SFO SFO SFO CFO SFOj j j j j j je e e e e eC D E eA B

Only add

MRC pilot

41

Comparison between the MRC pilot and the original pilot addition

Original pilot multiplication 21 7 7 21

, , ,CFO SFO CFO SFO CFO SFO CFO SFOj j j jA B Ce De e e

4 21 7 7 21CFO SFO SFO SFO SFOjA eBCD

After mutual multiplying

42

Channel A

MRC pilot 4 pilot multiplying

43