Nghin cu v thc thi kh ting vng m hc trn

kit thi gian thc TMS320C6713 DSK.

Subtitle as needed (paper subtitle)

Authors Name/s per 1st Affiliation (Author)

Authors Name/s per 2nd Affiliation (Author)

line 1 (of Affiliation): dept. name of organization

line 2-name of organization, acronyms acceptable
line 3-City, Country
line 4-e-mail address if desired

line 1 (of Affiliation): dept. name of organization

line 2-name of organization, acronyms acceptable
line 3-City, Country
line 4-e-mail address if desired

AbstractThis electronic document is a live template and

already defines the components of your paper [title, text, heads,
etc.] in its style sheet. *CRITICAL: Do Not Use Symbols, Special
Characters, or Math in Paper Title or Abstract. (Abstract)

II.

KH TING VNG M HC V THUT TON THCH NGHI

A. Kh ting vng m hc

Keywordscomponent; formatting; style; styling; insert (key

words)

I.

INTRODUCTION (HEADING 1)

Hin nay, x l tn hiu s tr thnh mt trong

nhng ng dng c bn trong nhiu lnh vc khc nhau nh:
x l tn hiu m thanh, x l nh (thu nhn v khi phc
nh), truyn d liu, kh xuyn knh, iu ch
Cc h thng truyn thng hin i ngy nay ang pht
trin nhm to ra qu trnh hi thoi thn thin hn, trng
hp mt phng hi ngh nu b nh hng bi nhiu th
chng trnh s khng cn s thuyt phc cho ngi theo di
do c thnh phn xen vo ting ni gc l cc ting vng
(echo). Ting vng sinh ra t rt nhiu nguyn nhn m i
hi mun lm gim hay trit tiu chc chn phi phn tch mi
tnh hung nh hng n ting ni. Do vn t ra l
phi loi b ting vng m hc cc tn hiu m thanh thu
c.
Bi bo nghin cu v b lc thch nghi (Adapter Filter),
cc gii thut nh LMS, VSLMS, MVSLMS, NVSLMS,
NLMS v VSNLMS kh ting vng m hc nhm nh gi
kt qu hot ng tng thut ton v la chn thut ton ti
u, nng cao cht lng thoi. T kt qu nghin cu tin
hnh thc thi thut ton ti u trn KIT TMS320C6713 DSK
trong qu trnh kh ting vng.

Fig 1. S khi b kh ting vng m hc

B kh ting vng v c bn l mt thit b c th pht

hin v loi b khi c ting vng xut hin trong ting ni gc
m ngi nghe cm nhn c. Ting vng l s kt hp
vi ting ni u xa v cc tn hiu m thanh b trlm nh
hng n cht lng ting ni gc.
B kh ting vng gm ba khi chc nng chnh [2]:
1) D thoi i (Double Talk Detector)
c s dng trong b kh ting vng cm nhn c
khi tn hiu u xa b nh hng li. Khi d thoi i phi
pht hin khi xy ra qu trnh thoi i v phn bit khi tn
hiu gn b nh hng ca tn hiu xa tc ng, nu b d
thoi i khng pht hin c c thoi i s l nguyn nhn
b lc thch nghi b phn k. V vy phi c b d thoi i
ng tin cy v nu kt thc qu trnh thoi i s tr li hot
ng bnh thng.
Vi x l tn hiu u xa, v l tn hiu u gn, H l p ng
xung, vi L l chiu di tn hiu ting vng.
.
H=[H 0 , H 1 , .... H L 1 ] T

d(n)=H T x(n)+v(n)
x(n)=[x(n),x(n-1),,x(n-L+1)] T
Tn hiu li s c thit lp vi cng thc:
e(n)=d(n)-H T x(n)
Tn hiu li ny s dng trong cc thut ton thch nghi
iu chnh cc tham s b lc.
2) B lc thch nghi (Adaptive Filter)
B lc thch nghi c to thnh t s c lng v b
tr. Ting vng c c lng trn ng nhn v xy dng
thut ton nhm kh i ting vng m hc. Thc hin php
chp p ng xung vi tn hiu ni nhn c lm i
tng hun luyn cho b tr thu c tn hiu sch nh mong
mun, ti thiu ha hm li. B kh ting vng l s hi t
ting vng trong lc x l thng qua cc b lc thch nghi. B
lc thch nghi s dng cc thut ton LMS, NLMS thc
hin kh ting vng m hc.
3) B x l phi tuyn tnh (Nonlinear Processor)
Mt b x l phi tuyn l mch x l tn hiu cui cng
lc b tn hiu cn st khi b lc thch nghi x l. Tn hiu
ting vng to nn bi tn hiu ting n gia tng, tn hiu b
bin dng c hy b hon ton hoc lm suy hao mt phn
ng k no .
li ca b x l phi tuyn l tham s ng c quan tm
nht trong qu trnh c tn hiu u vo tc ng, kch khi b
x l phi tuyn.

Fig 2 B x l phi tuyn tnh.

li khng bao gi ln hn 1 v c iu khin bi

mc tn hiu vo, khi mc tn hiu vo cao hn mc
ngng th vn hot ng bnh thng v li c khuch
i. Nhng khi mc tn hiu vo thp hn gi tr ngng th
li gim xung, tn hiu chm hn v c loi b hon
ton ty thuc vo tn hiu ngng.
B. Thut ton thch nghi kh ting vng m hc
Cc b lc ly mu cc thi im khc nhau nn b lc
phi thch nghi tng thi im. Trong thc t c nhiu thut
ton phi hot ng rt tt cho s thch nghi ca cc b lc.
Mun kh nhiu th thut ton LMS c s dng u tin v
tip sau pht trin ln thut ton VSLMS, MVSLMS,
NVSLMS, NLMS v VSNLMS v qu trnh tnh ton. L do
cho iu ny l cc th tc tnh ton ca chng l tng i
nh v d dng thc hin trong ng dng kh ting vng.

Fig 3. S tng qut ca qu trnh kh ting vng.

B lc thch nghi c th l IIR hoc FIR, b lc FIR l
trng hp c bit ca b lc IIR khi thc hin b nhnh hi
tip. Hm truyn t ca b lc FIR khng c im cc, khi
h thng b lc lun lun n nh. Nhng i vi IIR c c
im cc v im khng, b lc IIR trong qu trnh thch nghi
c th cp nht n cc gi tr nm ngoi vng trn n v,
lm cho b lc khng n nh [8]. V thc t khng th thit
k c b lc IIR. lc b tn hiu ting vng cng vi s
xen vo ca tn hiu u xa cn yu cu mt b lc n gin,
n nh, d dng thit k phc v qu trnh lc tt nht, cc b
lc ang c s dng trong kh ting vng m hc (AEC)
thng l chn b lc thch nghi FIR.
1) Thut ton LMS
a) Gii thiu
Thut ton LMS l thut ton thng c ng dng rng
ri cho cc b lc thch nghi, bi v cc l do sau y [9]:
L gii thut c bn u tin.
Do yu cu t php ton nn tnh ton n gin.
Trong thc t n lm vic rt tt, tnh n nh cao.
Vic cp nht cc h s rt n gin nn n thch
nghi lin tc cho b lc.
Cc h s hay trng s ca b lc FIR thch nghi c th
hiu chnh da trn mt mi trng thay i nh mt tn hiu
vo. Cng c th s dng b lc IIR thch nghi, nhng cc
im cc ca mt b lc IIR trong qu trnh thch nghi c th
cp nht n cc gi tr nm ngoi vng trn n v, lm cho
b lc khng n nh.
Chc nng chuyn giao ca b lc l s c lng, hm
chuyn giao ln lt c s dng trong tnh ton c lng
cho ting vng, h s nhnh c cp nhp th hin qua cc
cng thc sau [4]:

x ( n) [ x (n) x ( n 1) x (n 2)...x (n N 1)]T

e( n ) d ( n ) y ( n )
y ( n) w T ( n)
Trong :
x(n): l vector u vo ca gi tr ng vo b tr theo thi gian

x ( n) [ x (n) x ( n 1) x (n 2)...x (n N 1)]T

w(n): l vector biu din cc h s ca vector trng s nhnh
b lc FIR thch nghi ti thi im n

d(n) : l tn hiu mong mun (desired signal) hay cn gi l

tn hiu b nh hng ting vng (echoed signal) c s dng
trong bi bo ny.
y(n): l tn hiu ng ra ca b lc thch nghi.
e(n): l sai s c lng c xem nh s sai khc gia tn
hiu d(n) v y(n).
T : l ma trn chuyn v.
: l tham s kch thc bc (step size) v l mt hng s
dng nh. Vic la chn gi tr thch hp l cn thit bi
n nh hng n hiu sut ca gii thut, tham s nh gi
tc hi t ca thut ton.
b) Qu trnh thc hin
Mi vng lp ca gii thut LMS cn thc hin ba bc
theo th t sau :
Bc 1 : Tnh ng ra ca b lc, y(n).
N 1

y ( n) w( n) x(n i ) wT ( n) x(n)

0.001

0.0175

1.0680

-1.2073

0.0015
0.002
0.004
0.006
0.008
0.01
0.02

0.0166
0.0160
0.0143
0.0131
0.0122
0.0115
>>>

1.5436
1.7664
2.4390
2.9576
3.3989
3.7892
<<<

-1.4855
-1.7149
-2.4181
-2.9737
-3.4472
-3.8659
>>>

Khi tin hnh thc nghim vi s thay i ca chiu di

b lc v tham s tng ng nh TABLE 1 v TABLE 2, kt
hp nghe m thanh tn hiu ra cng vi thi gian thut ton x
l, MSE, ERLE tng i chn = 0.001. S la chn tt
nht l thi gian x l cc php tnh nhanh nht, MSE li bnh
phng trung bnh v ERLE tha mn cht lng u ra m
thanh nghe tt, tuy vy nhng bin tn hiu b. Khi chn
tham s ti u cn phi kt hp cc tiu chun nh gi trung
bnh gia cc h s cht lng v cht lng m thanh.
Tin hnh m phng thut ton LMS trn Matlab vi =
0.001.

MSE

ERLE [dB]

AV [dB]

0.001

0.0223

0.1911

-0.1531

0.0015
0.002
0.004
0.006
0.008
0.01
0.02
0.05

0.0221
0.0219
0.0211
0.0204
0.0198
0.0193
0.0173
>>>

0.2593
0.3234
0.5579
0.7702
0.9669
1.1508
1.9293
<<<

-0.2100
-0.2638
-0.4689
-0.6619
-0.8485
-1.0241
-1.7725
>>>

TABLE 2 BNG GI TR THC NGHIM CA THUT TON LMS KHI THAY I

THAM S VI CHIU DI B LC M = 2000.
Kt qu
Tham s
MSE

ERLE [dB]

AV [dB]

Amplitude
Amplitude

Input Signal

1
0
-1

0.5

Iterations

1.5

2.5
4

x 10

Echoed Signal

1
0
-1

0.5

Iterations

1.5

2.5
4

x 10

Output Signal

0.5
0
-0.5

0.5

Iterations

1.5

2.5
4

x 10

LMS Algorithm

0.08

MSE

0.06
0.04
0.02
0

0.5

Iterations

1.5

2.5
4

x 10

ERLE [dB]

TABLE 1 BNG GI TR THC NGHIM CA THUT TON LMS KHI THAY I

THAM S VI CHIU DI B LC M = 1000 .
Kt qu
Tham s

Amplitude

i 0

(2.1)
Bc 2 : c lng sai s, e(n).
(2.2)
Bc 3 : Cp nht trng s nhnh ca b lc cho
vng lp k tip, w(n+1).
(2.3)
L do chnh cho s ng dng ph bin ca cc gii thut
LMS trong lc thch nghi l do s tnh ton n gin. i vi
mi vng lp, gii thut LMS yu cu 2N php cng v 2N+1
php nhn (N tnh ton ng ra, y(n), mt cho 2e(n) v mt
N b sung cho php nhn v hng ca vector).
c) Kt qu m phng
Vi tn hiu ting ni sch a vo b lc cng vi p
ng xung c x l bng thut ton LMS tm c lng sai
s lm i tng hun luyn cho b lc thch nghi, tip tc
cp nhp trng s nhnh. Thc hin m phng, thc nghim
bng cch thay i tham s kch thc bc tm ra li trung
bnh bnh phng nh nht (MSE), h s tn hao ngc
(ERLE) v suy hao trung bnh (AV) cao nht, kt hp ng
thi vi bin tn hiu u ra v cht lng m thanh nghe
c khi tin hnh qu trnh thc nghim.

3
2
1
0

0.2

0.4

0.6

0.8

1
1.2
Iterations

1.4

1.6

1.8

2
4

x 10

Fig 3. Kt qu ng ra, MSE v ERLE ca thut ton LMS.

2) Thut ton NLMS

a) Gii thiu
Mt trong nhng nhc im chnh ca gii thut LMS l
tham s c nh i vi mi vng lp. iu ny yu cu cc s

liu thng k ca tn hiu ng vo trc khi bt u thc hin

lc thch nghi. Trong thc t iu ny kh t c. Ngay c
khi chng ta gi nh tn hiu ng vo h thng kh echo thch
nghi ch l tn hiu ting ni (speech) th vn cn nhiu yu t
nh nng lng u vo tn hiu v bin s nh hng n
hiu sut ca gii thut. Chnh v vy, hnh thnh nn gii
thut NLMS. NLMS l gii thut m rng ca gii thut LMS,
khc phc c nhc im ca gii thut LMS bng cch
chn gi tr kch thc bc khc nhau, i vi mi vng lp
ca gii thut. Kch thc bc ny t l nghch vi tng nng
lng trung bnh ti cc ng vo nhnh b lc.
Phng trnh quy ca gii thut NLMS c th hin
[2]:

1
w( n 1) w(n) T
e( n ) x ( n )
x ( n) x ( n)
w( n 1) : gi tr trng s nhnh k tip ca b lc

thch nghi.

w(n) : gi tr trng s nhnh hin ti ca b lc

thch nghi.
suy ra gii thut NLMS chng ta i xem xt quy
LMS cho vic la chn mt tham s kch thc bc thay i
l (n). Tham s ny c chn cc gi tr sai s l
e (n) s c ti thiu bng cch s dng cc trng s
nhnh b lc c cp nht l w( n 1) , v vector ng vo
hin ti l x(n).

w(n 1) w( n) 2 (n)e(n) x (n)

e ( n) d (n) wT (n 1) x( n)
vo

phng

trnh

(1.18)

ta
2

(n) , chng ta tm cc gi tr (n) bng cch gn

e (n) =0, nh vy kt qu thu c:
1
( n)
T
2 x ( n) x ( n)
Gi tr trng s nhnh k tip ca b lc thch nghi
quy NLMS

1
e( n ) x ( n )
x ( n ) x ( n)
T

b) Qu trnh thc hin

Khi tham s kch thc bc c chn da trn cc gi tr
u vo hin ti, gii thut NLMS cho thy c s n nh,
n gin v hiu qu hn so vi gii thut LMS. Chnh iu
ny lm cho gii thut NLMS l mt s la chn l tng cho
h thng kh echo m hc thi gian thc.
Bc 1 : Tnh ng ra ca b lc thch nghi, y(n).
N 1

y ( n) w( n) x(n i ) wT ( n) x(n)
i 0

(2.17)

thut NLMS, n nm trong phm vi ca 0 1 ,

gi tr l mt hng s dng nh khc 0 khi cc gi
tr vector ng vo bng 0.
~

w(n 1) w(n) T
e( n ) x ( n )
x ( n) x ( n)
(2.19)

Vi =1, = 0:
~

( n)
T
2( x (n) x(n) )

Bc 2 : Tnh c lng sai s (error signal), e(n).

(2.20)

Bc 4 : Cp nht trng s b lc chun b cho

vng lp k tip, w(n+1):
(2.21)
c) M phng
m thanh u ra ca thut ton NLMS c bin ln.
Khi kt hp chn ly thch hp khng nhng nh gi da
trn l thuyt m cn s dng nghe cht lng m thanh
u ra.
TABLE 3 BNG GI TR THC NGHIM CA THUT TON
NLMS KHI THAY I THAM S VI CHIU DI B LC M =
1000
MSE

ERLE
[dB]

AV [dB]

0.001

0.0230

0.0229

-0.0132

0.01
0.03
0.04
0.045
0.05
0.1
0.15
0.2

0.0226
0.0219
0.0214
0.0215
0.0214
0.0203
0.0194
0.0185

0.1411
0.03081
0.0808
0.4153
0.4488
0.7445
1.0057
1.2518

-0.0784
-0.1804
-0.2336
-0.2600
-0.2864
-0.5320
-1.6632
-0.9846

c:

ti thiu ha hm bnh phng (e ( n)) li i vi

w( n 1) w(n)

Bc 3: Tnh gi tr kch thc bc, tham s

c s dng nhm thay i tc hi t ca gii

Tham s

e ( n) (1 2 (n) x T ( n) x (n))e(n)

Vi:

Thay

(2.18)

TABLE 4

BNG GI TR THC NGHIM CA THUT TON NLMS KHI

M = 2000

THAY I THAM S VI CHIU DI B LC

Tham s

MSE

ERLE
[dB]

AV [dB]

0.001

0.0226

0.0841

-0.0764

0.01
0.03
0.04
0.045
0.05
0.1
0.15
0.2

0.0202
0.0183
0.0176
0.0174
0.0171
0.0153
0.0139
0.0129

0.5329
0.9739
1.1300
1.2011
1.2685
1.8219
2.2651
2.2683

-0.4944
-0.9316
-1.0936
-1.1676
-1.2384
-1.8092
-2.2502
-2.6494

Khi thc thi, gi tr tham s c th ln n gi tr 1 nhng

kt hp vi cht lng m thanh u ra vi gi tr tham s
= 0.2 th tn hiu khng kh c ting vng. nh gi MSE,

ERLE v AV cng kt hp cht lng m thanh u ra chn

tham s ti u cho thut ton NLMS =0.05.
Kt qu m phng trn Matlab:
Input Signal

Amplitude

1
0
-1

0.5

Amplitude

Iterations

1.5

2.5
4

x 10

Echoed Signal

1
0
-1

0.5

Iterations

1.5

2.5
4

x 10

Output Signal

0.5
Amplitude

III. THC THI THUT TON KH TING VNG M HC

TRN KIT TMS320C6713 DSK

0
-0.5

0.5

Iterations

1.5

Fig5. Kit TMS320C6713 DSK

Trn c s nghin cu v thc nghim trn i vi cc
thut ton LMS, NLMS s i so snh, nh gi la chn ra
thut ton tt nht thc thi thut ton trn KIT
TMS320C6713 DSK Fig 5. ng thi xy dng Simulink trn
phn mm Matlab tin hnh thc thi kh ting vng trn
KIT TMS320C6713 DSK.

2.5
4

x 10

NLMS Algorithm

0.08

MSE

0.06
0.04
0.02
0

0.5

Iterations

1.5

2.5
4

x 10

1) Qu trnh thc hin

ERLE [dB]

Begin

2
1

Xy dng lu v m phng cc thut ton trn Matlab

0
-1

0.2

0.4

0.6

0.8

1
1.2
Iterations

1.4

1.6

1.8

2
4

x 10

Fig 4 Kt qu ng ra, MSE v ERLE ca thut ton NLMS

Bi bo nghin cu, xy dng v m phng thut ton
thch nghi LMS, NLMS. Tin hnh qu trnh thc nghim
bng cch thay i cc tham s nh hng thu c cc
yu t lin quan n nh gi cht lng thut ton. Trn c
s cc gii thut , i m phng trn MatLab thu c m
thanh tn hiu ra v cc h s nh gi cht lng thut ton.
Kt qu thu c l cc hnh nh, th v cc file m thanh
nhm hiu r qu trnh tnh ton, cp nhp, kh ting vng ca
tng thut ton. Chn cc tham s ti u cho tng thut ton
lm tin cho qu trnh so snh, la chn thut ton ti u
nht thc thi trn KIT thi gian thc

Thc nghim ti u tham s cho cc thut ton

So snh, nh gi, la chn thut ton ti u nht

Xy dng m hnh kh ting vng trn Simulink

Thit lp giao tip gia my tnh v KIT DSP

Tin hnh kh ting vng trn KIT TMS320C6713 DSK

End

2) Thut ton LMS

Begin

Begin
c cc tham s , M,
itns, v khi to w(0), x(0)

c cc tham s , M,
itns, v khi to w(0), x(0)

n=1

n=1

c tn hiu vo, x(n) v tn hiu

b echo, d(n)

c tn hiu vo, x(n) v tn hiu

b echo, d(n)

Tnh ng ra b lc, y(n)

y(n) = wTx(n)

Tnh ng ra b lc, y(n)

y(n) = wTx(n)

c lng sai s, e(n)

e(n) = d(n) y(n)

c lng sai s, e(n)

e(n) = d(n) y(n)

Cp nht trng s nhnh, w(n+1)

w(n+1) = w(n) + 2e(n)x(n)

n = itns?

Cp nht trng s nhnh, w(n+1)

w(n+1) = w(n) + 2e(n)x(n)

n=n+1

No

No

n = itns?
Yes

Yes
end

End

Fig 6. Lu thut ton gii thut LMS

3) Thut ton NLMS

n=n+1

Fig 7 lu thut ton NLMS

4) La chn thut ton ti u thc thi trn KIT
Cc tham s ti u ca cc thut ton th hin trong
TABLE 5 nhm tin n m phng so snh trn Matlab. Vi
quy nh M l chiu di b lc, p ng xung l file c ghi
m trong phng hc, vi vng lp l chiu di tn hiu vo, tn
hiu u vo qua thc nghim vi 20 file m thanh sch
chn c file m thanh tn hiu s 7 (sp7), tn hiu mong
mun (tn hiu b ting vng) c chp bi tn hiu u vo
v p ng xung thc t. Tin hnh thc nghim cc tham s
c lin quan n s thay i kch thc bc, thu c tn
hiu u ra nh mong mun.
TABLE 5 Bng gi tr cc tham s ti u cc thut ton
Thut
ton

Tham s ti u

MSE

ERLE[dB]

AV[dB]

LMS

M=1000, =0.001
M=2000, =0.001

0.0223
0.0175

0.1911
1.0680

-0.1531
-1.2073

NLMS

M= 1000, =0.05
M= 2000, =0.05

0.0214
0.0171

0.4488
1.2685

-0.2864
-1.8092

Da vo bng trn cho thy khi tng chiu di b lc th

cc h s cht lng MSE, ERLE v AV ci thin ng k.

Thut ton LMS chn chun ha khi kt hp nghe m

thanh tn hiu ra cng vi thi gian thut ton x l, MSE,
ERLE tng i chn = 0.001. S la chn tt nht s
m bo tha mn cc h s cht lng MSE, ERLE, AV v
cht lng u ra tt.
Thut ton NLMS c m thanh thu c t cht lng
tt cng vi bin tn hiu cao, ngoi ra m bo cc gi tr
MSE, ERLE v AV. Vic chn thch hp khng nhng nh
gi da trn l thuyt m cn cm nhn nghe cht lng m
thanh u ra tt hay xu chn tham s ti u cho thut ton
NLMS =0.05.
Fig 9. Xy dng khi kh ting vng dng NLMS

Fig 8 So snh ng ra gia LMS, NLMS vi chiu di b lc

1000
Da theo kt qu m phng trn Hnh 4.7 v Hnh 4.8 cng
vi cht lng m thanh u ra, chiu di b lc 1000.
TABLE 5 Bng nh gi hot ng ca cc thut ton
Qua TABLE 5 tm tt v hot ng ca cc thut ton thch
nghi. V mt ton hc, nh gi c hiu sut ca mi
thut ton c th s dng cc thng s MSE, ERLE, AV. Tham
s MSE ca thut ton NLMS c gi tr b nht, ERLE c gi
tr tng i cao so vi cc thut ton cn li, thut ton
NLMS m bo h s cht lng tt. S b nhn ca NLMS
c tnh kh thi trong vic thc thi trn KIT. Thut ton NLMS
m bo c bin tn hiu ra v cht lng m thanh tt.
Nh vy NLMS l s la chn tt nht cho h thng kh ting
vng theo thi gian thc. Qu trnh kh ting vng m hc
theo thi gian thc c pht trin thnh cng vi thut ton
NLMS trn KIT TMS320C6713 DSK. Thut ton ny khi kt
hp cc h s cht lng (MSE, ERLE, AV) v m thanh u
ra, c kh nng kh c ting vng vi thi gian tr ln
n 75ms, khong cch tng i ca ting vng n cc mi
trng cch nhau 12m [5].
5) Thc thi thut ton NLMS trn Kit
Thc thi kh ting vng m hc trn KIT TMS320C6713
c xy dng bng phng php s dng Matlab Simulink.
Qu trnh c thc hin nh sau:

Fig 10. Khi kh ting vng m hc chi tit dng NLMS

Tin hnh qu trnh a ting ni vo Microphone ly
tn hiu ra theo cc switch thit lp sn trong khi hot ng
nh sau: nu khng nhn Switch no l tn hiu u ra ca
khi echo.
Nhn Switch 1: Tn hiu ra c kh vng (tn hiu
Thut MSE
ERLE
B
nh gi
ton
nhn
LMS

0.0223

0.1911

2N+1

NLMS

0.0214

0.4488

3N+1

n gin nht thc hin

v n nh khi thng s
kch thc bc c
chn mt cch thch hp.
Cht lng m thanh ng
ra tt nhng khng cp
nhp tham s qua tng
vng lp.
Thc hin n gin v
hiu sut cao. Cp nhp
kch thc bc lin tc
qua tng vng lp. Hot
ng n nh. Cht lng
m thanh tt v bin tn
hiu cao. y l s la
chn tt nht thc thi
trn KIT thi gian thc.

sch)
Nhn Switch 2: Tn hiu ra ca khi tng cng echo
Nhn Switch 1 v Switch 2: Tn hiu u vo
Nhn Switch 3: Tn hiu li

[1]

[2]
[3]

G. Eason, B. Noble, and I.N. Sneddon, On certain integrals of

Lipschitz-Hankel type involving products of Bessel functions, Phil.
Trans. Roy. Soc. London, vol. A247, pp. 529-551, April 1955.
(references)
J. Clerk Maxwell, A Treatise on Electricity and Magnetism, 3rd ed., vol.
2. Oxford: Clarendon, 1892, pp.68-73.
I.S. Jacobs and C.P. Bean, Fine particles, thin films and exchange
anisotropy, in Magnetism, vol. III, G.T. Rado and H. Suhl, Eds. New
York: Academic, 1963, pp. 271-350.

[4]
[5]
[6]

[7]

K. Elissa, Title of paper if known, unpublished.

R. Nicole, Title of paper with only first word capitalized, J. Name
Stand. Abbrev., in press.
Y. Yorozu, M. Hirano, K. Oka, and Y. Tagawa, Electron spectroscopy
studies on magneto-optical media and plastic substrate interface, IEEE
Transl. J. Magn. Japan, vol. 2, pp. 740-741, August 1987 [Digests 9th
Annual Conf. Magnetics Japan, p. 301, 1982].
M. Young, The Technical Writers Handbook. Mill Valley, CA:
University Science, 1989.