2008 (matlab) , , (ec) - tishreen.edu.sy file٢٢٩ tishreen university journal for research and...

٢٢٩

Tishreen University Journal for Research and Scientific Studies - Engineering Sciences Series Vol. (30) No. (2) 2008

2008

��

(MATLAB)

(EC),,

Generated by Foxit PDF Creator © Foxit Softwarehttp://www.foxitsoftware.com For evaluation only.

٢٣٠

Tishreen University Journal for Research and Scientific Studies - Engineering Sciences Series Vol. (30) No. (2) 2008

Studying and Handling the Phenomenon of Echo in Communication Systems

Dr. Haytham Al-Radwan

Dr. Sadek Ali** Faifaa Mikaiel***

(Received 6 / 1 / 2008. Accepted 18 / 3 / 2008)

� ABSTRACT �

The phenomenon of echo in wired and wireless communication is considered to be

one of the most important obstacles that affects the quality of the signal and system performance. This research aims to study the phenomenon of echo, its reasons, and its mechanisms. Bearing the outcome in mind, a method is to be suggested to overcome these obstacles. The echo and the echo canceller were simulated by MATLAB using NLMS to suppress the echo of the communication channel. This paper offers effective suggestions to overcome the echo signal and therefore ensure the quality of communication.

Keywords: Satellite communication, Echo, Echo canceller, Hybrid, Acoustic, NLMS.

* Associate Professor, Department of communication Engineering, Faculty of Mechanical and

Electrical Engineering, Tishreen University, Lattakia, Syria.**Associate Professor, Department of Communication Engineering, Faculty of Mechanical and Electrical Engineering, Tishreen University, Lattakia , Syria. ***Postgraduate Student, Department of Communication Engineering, Faculty of Mechanical and Electrical Engineering, Tishreen University, Lattakia, Syria.


Tishreen University Journal. Eng. Sciences Series

٢٣١

,(space segment)

(ground segment)

(computers)(mobiles) (PSTN)

(Public Switched Telephone Network)

(MATLAB)

20062007


٢٣٢

-1

(30ms)

(30 ms)

[1]

(500 ms)

-1-1



٢٣٣

(186,000 miles per second)

(23,800mile)

(250 ms)

[1]

-1-2

.

2-wire(hybrid)

4-wire

(Hybrid Echo)

(Acoustic Echo)

[1]

-1-3

(PSTN)

(2-wire) (4-wire)

1(2-wire)

(4-wire)

[2]

1 2 3 6 4 5 7

المقسم المقسم


٢٣٤

1

2

3

4

5

6

7

-1-4

2

[2]

.(Echo Suppressors)

(Echo Canceller)

سماعة المشترك األول

voice

الثانيسماعة

voice



٢٣٥

-1(Echo Suppressors)

.

[1]

-2(Echo Canceller)

80%90%

Double Talk Detector (DTD)

(Non Linear Processor)(NLP

[1] [3]

(AF) (Adaptive Filter)

. (DTD)(Double Talk Detector)

.(NLP)(Non-Liner Processor)

(3)


٢٣٦

3

x(n)(FES)Far-end speaker

v(n)(NES)Near-end speaker

y(n)

d(n)

:e(n)

ĥ(n)

h(n)

:ŷ(n)

3 x(n)

ĥŷ(n)

x(n)

y(n)

(echo path)

d(n)

d(n)

-1:

(4)

+

x(n)

From Near-end speaker

Adaptive Filter

ĥ

Echo path h

Σ Σ

From Far-end speaker

d(n)=y(n)+v(n) v(n)

y(n)

e(n) +

+ -

ŷ(n)

To Far-end speaker



٢٣٧

4

:(4)

Ŷ(n)

Ŷ(n)=x(n)*h0(n)+x(n-1)*h1(n)+……….+x(n-L+1)*hL-1 (1)

(2)

1

0)()(

L

i i inxnh ŷ(n)

(n)= h(n)Tx(n) (3)

x(n)

(h)

h0 = [h0(0), h0(1),…. h0(N-1)] (4)

h0

N

x(n)

x(n)=[x(n),x(n-1),…x(n-N+1)]T (5)

d(n

d(n)=x(n)T h0+v(n), n=0,1,2,…..,L-1 (6)

,n=0,1,2,…..,L-1 (7)

1

0

0 )()()(N

i

nvinxih d(n)=x(n)*h0(n)+v(n)=

x(n)T .x(n)

L

(3)y(n)v(n)

ŷ(n)

e(n)=d(n)- ŷ(n) (8) e(n)= y(n)+v(n)- ŷ(n) (9)

X(n) X(n-1)

X(n-2) X(n-L+1)

h0(n) h1(n) h2(n) hL-1(n)

z-1

z-1

z-1

ŷ(n)

Σ Σ Σ


٢٣٨

e(n)=v(n)y(n)-

ŷ(n)

(Normalized Least Mean Square) (NLMS).[4,3]

(10) J=| e(n)|2=| d(n)- ŷ(n)|2

J

[3]

(11)

h(n) n

δδ<<1

µ0µ2 [3]

-2

(5)

5

)()()()(

)()1( nxnenxnx

nhnhT

From Near-end speaker

Adaptive

Filter ĥ

Echo path

h

Σ Σ

From Far-end speaker x(n)

d(n) v(n)

y(n)

e(n)

+ + -

+

ŷ(n)

To Far-end speaker

DTD

Open during double talke

NLP



٢٣٩

(Geigel Algorithm)

[6,5]

N

Td(n)

|}1) N -x(t | , ... |,x(t)max{| (12)

α

(13)

T T=26dB2T=

3dB

[6,5]

-3

(5)

(6)

(6)NLP

-4(Echo Return Loss)(ERL)

6dB [7]

-5(Echo Return Loss Enhancement )(ERLE)

y(n)e(n)

dB

Td(n)

|}1) N -x(t | , ... |,x(t)max{|

level detector Gain control output

input


٢٤٠

(14)

)(nepe (n)

)(nypy (n)

(7)

All signal passed

start

FES signal x(n)

NES signal v(n)

ECHO signal y(n)

Microphone signal d(n)=V(n)+y(n)

DTD

The signal go to the NLMS filter

loop

α >T

yes

no

Subtract the good signal from error

signal

The signal go to the FES

The signal go to the NLP

)(

)(

10log10ne

ny

p

pERLE



٢٤١

MATLAB

(command window)

(Far End Speech)(FES)(Near End Speech)(NES)

[8]

(8)

0 5 10 15 20 25 30-4

-3

-2

-1

0

1

2

3

4

Time [sec]

Ampli

tude

Near-End Speech Signal

8v(n)

(9)

0 5 10 15 20 25 30-4

-3

-2

-1

0

1

2

3

4

Time [sec]

Ampli

tude

Far-End Echoed Speech Signal

9y(n)


٢٤٢

(10)

0 0.5 1 1.5 2

x 104

-4

-3

-2

-1

0

1

2

3

4Microphone Signal

Time [sec]

Ampli

tude

10d(n)

>02>

(ERLE)

fs=8KHZv(n)

0.025=

0 5 10 15 20 25 30-4

-2

0

2

4

Time [sec]

Ampl

itude

Output of Acoustic Echo Canceller

u = 0.025

11e(n)



٢٤٣

0 5 10 15 20 25 300

5

10

15

20

25

30

35

40

Time [sec]

ER

LE1

[dB

]Echo Return Loss Enhancement1

= 0.025

12ERLE

0.04=

0 5 10 15 20 25 30-4

-3

-2

-1

0

1

2

3

4

Time [sec]

Ampl

itude

Output of Acoustic Echo Canceller

u = 0.04

13e(n)

Time(sec)


٢٤٤

0 5 10 15 20 25 300

5

10

15

20

25

30

35

40

Time [sec]

ER

LE2

[dB

]

Echo Return Loss Enhancement1

= 0.04

14ERLE

v(n)=0

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

x 105

-100

-50

0

50

100

150

200Erorr signal

= 0.025

15e(n)

NLMS

Time(sec)

amplitude



٢٤٥

0 20 40 60 80 100 120 140 160 180 200-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1filter coefficients

= 0.025

15)

0 0.5 1 1.5 2

x 105

-100

-50

0

50

100

150

200Erorr signal

time(sec)

ampli

tude

= 0.04

16e(n)

Time(sec)

magnitude


٢٤٦

0 50 100 150 200-1

-0.5

0

0.5

1filter coefficients

frequency

mag

nitud

e

= 0.04

17

MATLAB

µ

µ

ERLE

ERLE(8,10)

ERLE6dB

ERLE6dB

PNLMS(Proportionate Normalized Least Mean Square)



٢٤٧

1- INTELSAT Digital Satellite Communication Technology. Revision 2,Aprile 1995.

2- WALLIN ,F. Combining Acoustic Echo Cancellation and Suppression, Linkoping

University, Sweden ,October 2003. pp,71. 3-International Telecommunication Union (ITU) "Digital network echo canceller

"Recommendation ITU-T,G.168. 4- FARHANG,B, Adaptive Filters: Theory and Applications, Chichester, England,

Wiley, 1998

5- BENESTY, D.R. MORGAN and J.H. CHO, “A New Class of Doubletalk Detectors Based on Cross-correlation”, IEEE Trans. Speech Audio Processing, vol. 8, pp. 168-172, March 2000.

6- BENESTY, T. GANSLER, D.R. MORGAN, M.M. SONDHI and S.L. GAY, “Advances

in Network and Acoustic Echo Cancellation”, Springer-Verlag, 2001. 7-International Telecommunication Union (ITU) "Application of the E-model :A planning

guide "Recommendation ITU-T,G.108. 8-“Matlab Real Time Workshop; User’s Guide,” The Mathworks Product Documentation. 5/7/2007,<http://www.mathworks.com/access/helpdesk/help/helpdesk.shtm>


٢٤٨

%the near-end speech signal v(n) savefile = 'test1.mat'; fs=8000; x=wavread('gg'); save(savefile, 'x','fs') plot(x) savefile = 'near-end speech. mat'; fs=8000; %g: wave file v=wavread('g'); p1 = audioplayer(v,fs); playblocking(p); save(savefile, 'v', 'p1','fs') n = 1:length(v); t = n/fs; plot(t,v); axis([0 33.5 -4 4]); xlabel('Time [sec]'); ylabel('Amplitude'); title('Near-End Speech Signal');

%the far-end speech signal x(n) savefile = 'Far end speech. mat'; fs=8000; %gg: wave file v=wavread('gg'); p8 = audioplayer(v,fs); playblocking(p8); save(savefile, 'v', 'p8','fs') load Far end speech x = x(1:length(x)); d= filter(H,1,x); t =(1:length(x))/fs; plot(t,d); axis([0 33.5 -4 4]); xlabel('Time [sec]'); ylabel('Amplitude'); title('Far-End Echoed Speech Signal'); set(gcf, 'Color', [1 1 1]) p2 = audioplayer(d,fs); playblocking(p2);

%Microphone Signal fff=length(v) s=dhat(1:length(v)); ss=0.001*randn(length(v),1); f=length(s) ffff=length(ss) g=s+v; d=ss+g; subplot(3,3,3) r=1:length(d);



٢٤٩

rr=r/fs; plot(r,d); axis([0 2*10^4 -4 4]); xlabel('Time [sec]'); ylabel('Amplitude'); title('Microphone Signal'); set(gcf, 'Color', [1 1 1]);

%NLMS algorithm fs = 8000; function [e,h]=nlms(µ,L,x,d,δ); % Normalized LMS % Call: % [e,h]=nlms(µ,M,x,d, δ ); % % Input arguments: % µ = step size, dim 1x1 % L = filter length, dim 1x1

% x = input signal, dim Nx1 % δ = constant, dim 1x1 % Output arguments: % e = estimation error, dim Nx1 % h = final filter coefficients, dim Mx1 %intial value 0 w=zeros(L,1); %input signal length N=length(u); %make sure that u and d are colon vectors x=x(:); d=d(:); %NLMS for n=L:N uvec=u(n:-1:n-M+1); e(n)=d(n)-w’*xvec; h=h+µ /(δ +xvec’*xvec)*xvec*conj(e(n));

end

%ERLE function [a]=erle(e,y); % Call: % erle=erle(e,y); % Input arguments: % e = residual echo, dim Nx1 % y = hybrid output signal, dim Nx1 % Output arguments: % erle = ERLE(n) in dB, n=1..N, dim Nx1 erle1=filter(Hd2,(e-v(1:length(e))).^2)./(filter(Hd2,dhat(1:length(e)).^2)); erledB1 = -10*log10(erle1); subplot(3,3,5); plot(t,erledB1); axis([0 33.5 0 40]); xlabel('Time [sec]'); ylabel('ERLE1 [dB]'); title('Echo Return Loss Enhancement1');


٢٥٠


2008 (matlab) , , (ec) - tishreen.edu.sy file٢٢٩ tishreen university journal for research and...

Documents