You are on page 1of 100

ECE551 Advanced Topics in Digital Signal Processing Homework #1 Due September 17, 2009

Review undergraduate material on discretetime signals and systems, and the DFT. Recommended reading: Proakis and Manolakis, chapters 16; or Oppenheim and Shafer, chapters 15, 8, 9, 11.1 and 11.2. et/2 : 0 t < 10 0 : else. Evaluate the resulting expression for p = 1, p = 2 and p . Problem 1. Compute the Lp norm of x(t) = Problem 2. (a) Evaluate the expressions (2t) x(t) dt, (t1/3 ) x(t) dt, and (t2 1) x(t) dt.

(b) Derive an expression for (f (t)) x(t) dt, where the function f (t) is arbitrary but continuous everywhere on the real line. Problem 3. A square pulse x(t) with amplitude and width equal to 1 is ltered through a LTI system with frequency response H(F ). Sketch the output x(t) for each of the three systems below, assuming that B = 10 Hz. Your sketches should be as realistic as possible and should clearly show (1) possible smoothing of the edges of the pulse, (2) overshoots, and (3) ripples in x(t). Compare the merits of the three systems with regard to criteria (1), (2) and (3).
F(Hz)

F(Hz)

sin ( PI * F / B)

PI * F / B

H(F)

H(F)

H(F)

B 1 -B 0

-B

-B

F(Hz)

Problem 4. The output y(t) of a linear time-invariant system with input x(t) = et u(t) is 2 given by y(t) = et u(t). Determine the impulse response of that system. Is the system BIBO stable? en/2 : 0 t < 10 0 : else. Evaluate the resulting expression for p = 1, p = 2 and p . Problem 5. Compute the lp norm of the sequence x(n) = Problem 6. How many continuous derivatives does the DTFT of x(n) =
1 1+|n|4

have?

Problem 7. Determine all possible stable sequences that have the following ztransform: 1 X(z) = (13z 1z)(12z 1 )2 . Indicate the location of the zeros and poles of X(z). Which one(s) of these sequences have a DTFT? Problem 8. Determine the ROC of X(z) =
n . n= sinc(n/2) z

Problem 9. The Fourier transform of some real, continuous-time signal xa (t) is zero at frequencies greater than B. The classical sampling theorem asserts that xa (t) may be recon1 structed from its samples taken at the rate 2B . Here we want to show that xa (t) may also be 1 reconstructed from its samples taken the rate T = B , so long as both the amplitude and the slope of xa (t) are measured at each sampling instant. Show that the reconstruction formula is

xa (t) =
n=

xa (nT ) a(t nT ) +
n=

xa (nT ) b(t nT ),

using the two interpolating functions a(t) = sinc2 (t/T ) and b(t) = t sinc2 (t/T ). Problem 10. Under what conditions can a signal xa (t) be reconstructed from measurements of its amplitude and all of its derivatives at time t = 0? Problem 11. In class we have discussed conditions for existence of the Fourier transform. We have also mentioned that it would be nice to extend the denition of the transform to signals such as pure sine waves, for which the Fourier integral does not converge. In this problem, you will show how to do this. Consider the constant signal x(t) 1 dened for < t < . We are tempted to write its Fourier transform as X(f ) = ej2f t dt. Unfortunately, such integrals cannot be interpreted in the usual sense, see footnote in Lecture 2 class notes, p. 5. Derive an expression for X(f ) using the following steps: 1. Dene a sequence of well-behaved signals x (t) such that lim x (t) = x(t). 2. Compute the Fourier transform X (f ) of x (t). 3. Discuss the behavior of X (f ) as . 4. Dene X(f ).

     ' S 7 ' !'   2dH5iFIFU&T%FU)2#`4gUU  X C   7    @ ! @ 7 1 { y  { y 9  %%i9HUFuUq%FF&UHFU$`2G#9G%FGy E&ys&#Rq%#AUX H1aHU&UHU#!EG#9qU)X#Sq%FVH%Veq)2&FqU))EUD$F)Ea#0s !    @ ` @  ' X ' 7   7 S '  ~ '  ' @ 9    ' ' @   ' @ 7 1 ! ' ` S  ! ' ! `  S  ! 1 ' 9   7  ' F'1AU#a$X&T2)FU0#HUUgUi%F 6e b f f f W &0"T W }"T W a" I f 9  B    '  ' 7  S ` 9   9   q%#0#!!&%7F70)2F&FTHUUgU7FDHUUgU(auG9#FlTUFTSF$XG'a%#XF2&(5 3    # uo24 z % g " }&Ts o#8%2%$i%FH%02U%&%F#AUC S  ` @ 7 7  ` @ `  7 9 u )"u6VI U @  U&UHFTSH%`uHF7F09HUFuUUgiI'#))e$TG#9F8)H%##!G9#FlTUFTSF$XG'a%#XF2&(5     ' 3  ' 3    $ uo24 z ! 1'  X 2)FFaa"E ! '  ` ' S  S F2aa"d  ' 3   1 Ve$')4I#TG$9Fa"s ' ! 9  7  ' )H##HUUgUa$F"e   B !  7  7S'    ' `    ` &FUH`IG%#Dd#!&%F YHAUV}U20%&gFU2& & sF FUIWWF4tQF v  X ! ' sI1HU)#'HFF)w W X'Q QDao~aC%700l D%`2U%&Y'o %2@#FDHUFuUG#9FlTUFTS#Xg'%#XF2&(5  @ ` X  ` 7 9  3    z uo24  `  S  9   X ' ! '  X C  ' ` X ' B X ` 9  !  ' 9  `  S B X ' U%&TDAHUUgU#&#BF' gs U)F&F8##'VU)AiU&Tga%%#2Ha

UNIVERSITY OF ILLINOIS AT URBANA CHAMPAIGN Department of Electrical and Computer Engineering ECE 551, Fall 2010 Issued: 8/24/2010 Not Due

! U$H u U u u o f % U f R0&G4  1 B U 4 ' 7   @ @ ` X IH&8#a0Gi)FU&%F2' ! F#H u U F uf u o f g U " hg U f 0HV 9  ' `  ~ B !  7  ' 7 H F  ' 7   @ @ ' UUa#%&%G%#Dd#!&%Fa8FITda)FU&%` s e ! ` ' S 9 ! 1  @ 7    ' @ 7 S ' X ' #!2F)H#2F%FoF)EVHHF  ' e H F   9    '  `  ~ 7    X  S    7  X !  1 X '   UH! fUIHd2UUoV%&%%Fq&y&F2&TQHT#S&%VS IHF&I#ys' eaFF)' y# II &R H Q

H S FI%T

 ` @ `  3 `  `  S X ' Q %2U%&G$9Fl"&%u#FI&Tw &R %`2@%#G9#FlTUFTS#X7F#0 HUFuU%FHU%H%Y#UqU#22G#9' Ur&FVe&(aTgFU# 3  9   7  X ' ! ` @    5   1 7  ! ! U#H u Q f Q U &V B !  7  1 B Q %%#Dd#!&%FuH&#( U &ViFD ' f ` &b )A%&Bo~ u f
E D

H F FITR

U 0&G4 D

H S b cwus q b Q UI%T pfX$xvtrpX i &R

f u E bcTx$$w Ry `uW s g c b` X hfedTaYW D

Q F &V D Q eG

CP 45

H F UIG

H1XHUT#S2%XHUUuF%F%X#F2&(5 ! @ 9   7  z uo24
B C

b I v & t & f d" t

fc % # ID(u o% # f t "7u % # f b 9s b A @ 8 f % # IE7u % # f t "g o% # f b G5s b 6 f 4 o% # Ig(u % # f t "(u % # f b "s b 3 2 1 % # b ) % # ' % # ! o&$Ig0u &$ f v (u &$ f b "s

    

! U#H u 4 5 4 1 u U 1 3 2af
0 )

U }0  7 F0

# !    ("'&

E D

H F FITR

U 0& D

5P aC
%

F D
# ! $"   

CP 45

H F UIG

H F FITR

5 P C

F D

CP 45 D

U 0& D UIG H F

9   AHUFuU} qB%#0#!!&%F%#XF2&(5  7   z uo24


 

9   ' ' S'  1' ' U X HUFuU8g0DFTsd)aDG4 2' %2U&owFIRy%E%#QyUIHdiF#0 HUUgU%Fo)FU }%$`4u2a%Fy#)  ` @ `   ' H F X '  ` @  ' H F 7 9   7  ' 7    ! ' x z  7  ' 7 u U U 0& F ' 7   @ @ ` ' 7 ' )FU&%F0#a)a(Ve F)2@$ HF%#X&Ti)FFeHs2AsdAFIG8 F%#X0%22A%F)2FGU&%F%E1 ' X   S  ' 7 ' 7  X 1  ' 9 H F  7 X ' 1 7  ' 7   @ @ `  ` ! U$H u U u u o f % U f R0&G4  1 B U 4  ' 7  '  ' @  '    @ @ 1 uH&#0Gi)F2VeiF)E#0%U&%` s s u u U f U 0 F 7  X ! X ' 1 '  ` @ 7 7  H F  ` @ `  X F0uWHUA#9$2F)F2%#%FH%0VFIR%2U%&2' H F  FIG F%`2@%$G7F0ug&b )F%&BoG# HUFuUaFs8U2&sFHFiTHIHF$!d%F#Hs@ s & ' f ` ~ 9   7     @  S `  ! '  7  S t U X '  U  Q  Q    7S  X ! 1 !  f V0& F 2EV0& S 2V F edV F eV DH%FH#Hs)$'%#XgdF d" v
    

7 7 ' FWiF#D)27

 ' 1 '  9  X  ! S X   ' 7 1 9 ` 7 S ` QgF)U#2F)0HU&##w g3 2 H#a)aiuVF f 3 7 t 3 f 3  S ` 7   )wf g"4u g3 F0yY g(%c g"W g3 &FT2%F%#XF2&(5
 

z uo24

! '   ' @    7 S ` 9  1 FU$#X2#aFHFUe$2dD s X '      9    ' S ! ' B 7     9  ' @ ! 7  9 7  ' 7 ' e G aAVU#2Uy%H%Fa%q) eHAud2g2Fe&G%F8#0)De ) X '  u  7 q5s Eg %G 4u %0 t os WT&Hf g }" W f f    ' 7  9  '  `    `  9 ' 7 7   9  ' @  7  9 ! ' ` H #F)FGFI&G}U)D(dA&e2&G0%FGHFeFIHGgH2qIFe&G2IH #A&#0%Frd2@ ' ! 7  '   ' `   ! 7  9 X X `  @ 9 S   '  ' '  !  X  1 `    @ @ U}e8g&s6 #Fe&AAH%&G&Tug 0U)FHFUe$g})R)#'#!&% FFA&U&` s

z uo24

9  7S'  HUUgU Ha U 0& F U F U F 0d 0&HV    @  S `  B ! ` 7    7S  X ! 1 !  HU&UHFTH%uHFq%#FHF%Fo0HFHH)#'%$XuF@  !  ` @   !S ' 7   ' 3 X ' ' !  '  ' 7 9  ! '  7  ' 7  %U#F%2#}F#')2Fe#'V)4I#TG#9F )H%#$HHH1 HUUgUQ#!Vdq%F)FU   '  9 U U$F)FAU`G0& F g)7FI'(HFe#VUU&TYFHUAHUUgUr(YUH%FH%~  '    ' S 7  S @  9     7   7S '  u u f u QUV b
    

UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN Department of Electrical and Computer Engineering ECE 551 Digital Signal Processing II Spring 2006

Problem Set 2
Issued: Thursday, January 26, 2005 Due: Thursday, February 2, 2005

Reading: Lecture notes 1-4, and Chapter 5 in DSP (Oppenheim et al.). Problem 2.1 Consider the causal discrete-time system with impulse response h[n] and system function H(z) = e1/z (a) Determine the unit pulse response h[n] of the system. (b) Determine whether or not H(z) corresponds to a minimum phase system.

Problem 2.2 One of the interesting and important properties of minimum-phase sequences is the minimumenergy delay property, i.e., of all the causal sequences having the same Fourier transform magnitude function, |H(ej )|, the quantity
n

E[n] =
m=0

|h[n]|2

is maximum for all n 0 when h[n] is the minimum-phase sequence. This result is proved as follows: Let hmp [n] be a minimum-phase sequence with z-transform Hmp (z). Furthermore, let zk be a zero of Hmp (z) so that we can express Hmp (z) as Hmp (z) = Q(z)(1 zk z 1 ), |zk | < 1,

where Q(z) is again minimum-phase. Now consider another sequence h[n] with z-transform H(z) such that |H(ej )| = |Hmp (ej )|
and such that H(z) has a zero at z = 1/zk instead of at zk .

(a) Express H(z) in terms of Q(z). (b) Express h[n] and hmp [n] in terms of the minimum-phase sequence q[n] that has z-transform Q(z).

Problem Set 2

(c) To compare the distribution of energy of the two sequences, show that
n n

[n] =
m=0

|hmp [m]|
m=0

|h[m]|2 = (1 |zk |2 )|q[n]|2 .

(d) Using the result of part (c), argue that


n n

|h[m]|
m=0 m=0

|hmp [m]|2 ,

for all n.

Problem 2.3 Consider the discrete-time LSI system with frequency response magnitude satisfying, |H(ej )|2 = + cos(), for || < . (a) Find the impulse response of the minimum phase system hmp [n] corresponding to this frequency response magnitude, i.e. |Hmp (ej )|2 = 5 + cos(), || < . 4 (b) Let B(z) = ( 1 + z 1 ). Show that |B(ej )|2 = 2
5 4

5 4

+ cos() as well.

(c) Determine the real-valued all-pass system G(z) such that B(z) = G(z)Hmp (z), where Hmp (z) is the system function for the minimum-phase system from part (a). (d) Determine
n= g[n]g[n

+ m].

Problem 2.4 Consider a discrete-time system with frequency response magnitude |H(ej )|2 which satises |H(ej )|2 = G(ej ) where G(ej ) is the discrete-time Fourier transform of the sequence g[n] and 1 5 1 g[n] = [n 1] + [n] [n + 1] 2 4 2 (a) Express g[n] in terms of h[n]. Provide as simple an expression as you can. (b) Can h[n] be determined directly from g[n]? If so, nd h[n]. If not, provide a mathematical description of all possible sequences h[n] consistent with this information. (c) If it is known that h[n] is a minimum-phase sequence, can h[n] be determined directly from g[n]? If so, nd h[n]. If not, provide a mathematical description of all possible sequences h[n] consistent with this information.

Problem 2.1

Problem 2.2

Problem 2.3

Problem 2.4 (a) G(ejw) = -0.5 e-jw + (5/4) 0.5 ejw = (1 0.5 e-jw)(1 0.5 ejw) Since, G(ejw) = |H(ejw)|2 = H(ejw) x H*(ejw) So therefore, g[n] = h[n] * h[-n]

TEBaXGDVUVSXDtet$XGtPhFIHaDPh$Q$E`tQqQ$q @ W ES G W A @ p b sC @ S A@ Y P@ p @E E@ E U G W A @ A pSE TBaXGWTVSVSdDQdCWQqejth$8 U S e p P C H E SU P C E E b p S @ G 7 |"V@tt$XGbthsFIHVaDVcaGTBX@cEthqq$tp a 7 m

G W A s b W SU @ G b C G P S Y C r @ E p b B$VXSDqP0oXSWhq$0tdPV$dh`F$thXs| x } v v x } { v }{ b G p A b @ G P @ E @ W U S G p AG p hmQty"joqByt7 mTdgtDqdPt$Qq0TBaXGD|VqDathS @ G b WC P b G p AG p S W b S A @ G US b @ b HE@G@ 9 r q$gddVthXsctDaqh0d@DX$$h7q$mV&cdYD0qdPIBaBc BB cFx 7 V BdVthds$qDaqhX@DXc$h7A p WC P b G p AG p S W b S P@ b s @EC p S Y r e P C r @E p b  P @ bG b A WE@ S G S P S e b @ W p A e P C SEEC P @@E QVXhi$FhVFtVywQ$qhdsmgVidH$dPFdCBVoVqqdYt0QadtgtVDT$$FtIB$t$G U S H p @ G b p C r @E p b  b P S S W C p pW C P b % C r @E p b  b P S @E C P bG Y P pU W @ Vt$q$0#yF$qhdsm6dP0oqc&aV7 |DVthds$yF$thXs|6dP0q$$F0thdS$`tthjQdCQte A p SE s e P C @ e p b P s C H @ P S A @E Y P @ p @EU G H @G  ! @ bG1 @G @E Y b e C b E @ b  qh$VtVqt$XGthFaDh$Q$j`tQqQ$$wjQaa$0yiidH$`aB$`$tBaXGdW|g ymu ~ }{ z x v P b C H E SU b W b G E C A b G b C G S SG S b W A @ p e @ b C G S @ P C Y @ W C @ @ G G p S hXS$GFIVqPVdC$$FDdD$dPV$tVIaoqdPVCD`q#QqdPV$tV7jVBQtdpVyaQq$qh7FC P bG C HE U b W GE C A W P S b p S b W A @ WE C WY p e P C p p p e P SG W b S A b b U hSd$FIcVSqdPqVdCb$$FXDhXUwotidPVdCDy`dcFQX@Bd| |itVd# |RB|Ry#tDad@DX$$h7'qXGTr G S P b W A @ p G S P U BVqR0PVdCDy`tBVq"f| m e P p p p P P b b e SY G C E@ e P p p S U e @E@ SY @E @ @Y P @ p @ W C b bE S @ G P C 2 tVC# mB|y#ghSthSX$XGtDPh`FD0Bqtt#qagQ$BV`Q$7i`tQtdQamVtdPhdscVq$gVFx c# | a| tVC"| m c# cYt0TVqB| tV# TV0DB`y`gdWVVq7 | 7 e P b E SU p ePC E SU G A @ Y @ W C E SU p V@tQVsqVtYcBaddPjPQB7@VotQ@qdpWVCFw`tQqQat$"VadDh$#7 mqdPyBX@caB$|V$Q`y$tq$ b p e @ P C E @G b @ @ C @ Y P @ p @ @ G U S S G p b e H SEU s b bEG @E U S @ Y SE A @ G P |  t ys d a p W C Y b Y @ A  e @ b W e P C b H E SU P C EG E b E p S  @ S @ Y P @ p @ W C @ E C @G S P @ e G @ ! VddWVBDXBQ7'QaXGIdHbttVDcV$tVcaTBX@qhqwahqw`tQqQa#VQ$IaVqQq7 |TB#& ymu ~ }{ z x v Y p ) b E @U C Y  p b @ P @ A A  5  9 b w E @G A C 2 s b e C @ E E C b E t$t$XG00BFtytdHQqtQ7tywdPiBatFtqqPtVQ$cFIdHcr5 uFsVsFrpwB@ttpVQCte$Eqd0TD"j p  } ussr p r PC pC eE p  } FVF7VfVhDVot$ctdTTooB } ~ }{ z x v d7 y|Dywu t hsVsFriqdPctytqBom70j#hgdD"yRvg s b E A  p n n l k i e f e x w u 2 s bE@@ b s P E@G p A H S 2 e PC WC bEGY W US G P@ HGEC A@ qdPcBBtdPhqrBaqDihgfDV#VBdYca`QX@'VTRQI$$FDB9 8 % 5 4 % 3 21 % % )  %   ! !      $07600('&$#" 

1 0.5

Input Signal x[n]

x[n]

0 -0.5 -1 0

50

100

150

200 n

250

300

350

400

20 15

Fourier Transform Magnitude of Input x[n]

|X(e^jw)|

10 5 0

0.2

0.4

0.6

0.8 1 1.2 1.4 Frequency Normalized by pi

1.6

1.8

Group Delay (Samples)

250 200 150 100 50 0 0 0.2 0.4 0.6

Group Delay of filter A

0.8 1 1.2 1.4 Frequency Normalized by pi

1.6

1.8

Frequency Response Magnitude of filter A 0

Magnitude(dB)

-50 -100 -150 -200 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Frequency Normalized by pi 1.6 1.8 2

}v DXgy" r

1 0.5

Possible Output y1[n]

y1[n]

0 -0.5 -1 -1.5 0 50 100 150 200 n 250 300 350 400

1 0.5

Possible Output y2[n]

y2[n]

0 -0.5 -1 -200

-150

-100

-50

0 n

50

100

150

200

1 0.5

Possible Output y3[n]

y3[n]

0 -0.5 -1 -1.5 0 50 100 150 200 n 250 300 350 400

1 0.5

Possible Output y4[n]

y4[n]

0 -0.5 -1 0

50

100

150

200 n

250

300

350

400

z }v tdgy t

1 0.5

Possible Output y5[n]

y5[n]

0 -0.5 -1 -200

-150

-100

-50

0 n

50

100

150

200

1 0.5

Possible Output y6[n]

y6[n]

0 -0.5 -1 0

50

100

150

200 n

250

300

350

400

1 0.5

Possible Output y7[n]

y7[n]

0 -0.5 -1 0

50

100

150

200 n

250

300

350

400

1 0.5

Possible Output y8[n]

y8[n]

0 -0.5 -1 0

50

100

150

200 n

250

300

350

400

}v td7y u

G $EFCDAHhScUgh HQ@aayaX@DX$chfcVQTVh VVXSDI$BB`d@hVXfoh E G W b S A Y C @ E SU E SU G W A SE @ 1 W S A C @ b 8 Y S P W b S A bG @E C C b @ b S Y @ G U S b otX@X$$h7dWdW$a$FRhdPQ`dYhqcq$"VjcdYt& @Y P@ p @ WC@E W@E p A V`tQqQa#VQ$XQ$qDIC dt7 7dh 'VcVatVcaFacBVddPwqTrh Vh$BBDVTQX@h 7"V0BtwiV$t$VtV@h b p p U S HE SU P CEG1 @E @ b @  C SE @ e P C W S A U S E @ H p P @ H C @ G C b G W S E @ eE S @ H C @ G U S e P C W C p C Y @ SG P S P S P b p S W % BqcV0IV$q$VtVV$tVB07a0otoqd'hQIdqa' RV s tVC cVVr x gBX@o e P w QaayaaVtDitHtdPIIftVQaaya$VDdAdWVjVC H @G @ C A1 H p b b H C e P C H @G C W C P U S @ e C Y C Y C e @ G P @ @ E A @ E @ P C Y P b G Y P p U H @ G P C G C G W C Y @ E @ P C E p S b G p VwtVB$VB&QaRQ$Q$tB$7VBhXS$`tthQaayaRVTFt$7WVBQThQBgatVTthVIdU$ath6ox $ DV0X@DF$ard'dh Ft$It$0$Bt$T$F'h 'thXS$`tthQa$aX@X$$h7RVoRh e P C W CG b G C G Y p @E @ G @E C P bG Y P pU H @G W b S A P C H S 3 C $ 7 dh 7 dh 7 m

W@ P S C p P bGY P p wXSQ70q$TVthdS$`tthU H @G W C P bG CE C b S W C p H @G s bG C P @ A H S Y G P bE C b1 @ bG E C @ W C b e @ e C Y C Y QaayaqVthXS$F$XG0j$dV'dhBIQaayajqP$F$tQ7ih`wdVdCcFFRdP`IdH$FQqdPdbf$XG0QqVB$VBwb T` x x 7 yyT` y0xx dh x T HE SU @ G U S P bG Y P pU H @G cVq$|VhXS$`tDQaa$ W C P bG CE C b H @G @ bG1@G@EY b e G P bE C b1@ bG E C @ W @ C A1 H p b b H1 P S P W CG C G C G @ S A A p VthXS$Fc$XG0(Qa$aidH$`aB$`$t#RVCcFyddP`iH$FQqdPdb|aVtDitHtdPIyhtrX@DF$a0Ft$'ah7tqq H @G @ C A1 H p b b H1 P S P C U S @ P S A @ E Y P @ p @ EU @ G U S @ e p b P s C H @ QaayaaVtDitHtdPIyhtImVaDh$Q$`DQqdQcwt$"VqD$XGthFIt$G XthtdP$F$tQ7ih`acVhcttFVgqt$$"QX@DVctdt$Qaaya$VtDtdHtPIyhq6VU W P S s bG C P @ A H SY SG Y C SE A A C P C e pG @ p H W SE A b G P H @G @ C A1 H p b b H1 P S P C E S E SG C P @ A H S Y G Y @U E @ A H @G @ E @ b W C G e P C W C p C Y C b C G S SG W b S A G S P b G VaF$tQ7ih``QcBQaayaacBVddPX@DF$atVmV$tVB6PV$tVaX@DX$chVqd ymu ~ }{ z x v

SG W C P bGE S A SE A @ G p H P @ G p H @G @E@ b @ G U S @ P S A @E @ W p A b @ 7 "aVthXS$$V7V$t7aD"7 6Qq$tQaayawaBVRdPq$"V$th7aQ$adtDIdHwq$G d7 tVqQaayag$VDXAdWVX@DF$$mV0aDh$Q$TadDDidHq$dtVQqdpVF1 VQ$d#7 " b e PC p H@G C WC W CG C US @ P S A@E @ W p A b @ G b e PC e@ WC WC@E b U s b P SC@E E p S @GCG WEC W 2 @ pE qqdPth$VQ$TqhVaF$aXFQX@gVqaG E B@VB@qPwEVS#pV@qpaGQ@idHb$B@iHh$#V@qcaGwyQCXVCdbRQ@iQaGF$$&qP0oXSWhq$Bq$Bt0iqdPIBaB ymu E G S p pE W G P H @ C G s b W SU @ G E @ G @ @ b H E @G @ 9 ~ }{ z x v g  WC S '7WdV0EVU  q  G C G @ p sE C p Y GE C A U S W p @E @ G s b Ft$qh$Foh$FD|VXGt$Qcwq$qdP$0qu g  7 q  0   y 7  x G C G S p@Y P@ p @ S G @ G U S sE@ P@ U S P bG p bEG b e @ G @E C A H SY S  Ft$(q$DQQ`tQqQagq$|VV$BqQmVhXS$qDXaadtwq$$FDih`"0t h  HE SU P CEG1 C G C G  @ Y P @ p @ @ C A1 H p b b H @ G U S HE @G b e P C @E A cVatVca`VDFt$7 q`DQqdQ$aVDDtdHtdPq$|VTcBadP"7 y6DV"7 0cQ$ty'0r US HE@G b @E A dh  V0IcBadPdh cQ$ty'0Fx G C U S e C @ G  d0F|VVQaatPb Ryx F$BBTVDdh0Ft$t$jDVC G C SE @ C C G C G Y p e P $ $ GC G Y p Ft$fctc dh cVatVa`f$XGm7 "`tQtdQaBq$VtVBqe$th`oTVaVttdHtdPIgdPVhFd0dh  $Bq HE SU P C EG1 b @ Y P @ p @ E @ G S P C E @ b P S Y S @ C A1 H p b b H b C s C b @E @ Fx   0`h  h x  C @E A @ P C Y @ G C G S U S SE @ C @ Vdh0$Q$ty`VBgTFt$I$dhd'VIcBBiw7 Wp ES @ Ep E P b YP@p C 1 Hp b b H C @ @ ! W BGX@DV$@VIcHBEq$$GqDTdhd cHVS$UtVcCaEG`1$XG`@tQqQa@a@VtDAitHtdPI77 BG#gB0XSdWhSU VQVoctdiXGt$QctTV`tQqQaaVtDDdHDdPIjq$d07 "Qq0gs dWVVtHyyIib C e @ SE A b W p @E b  @ Y P @ p @ @ C A1 H p b b H @ G b P @ W C E SU H p b C H g 7 " 7 & bG P C p @ G p P bG Y P pU @ e p b P s C dG$RVtq$Dp a hdS$`tthqt$XGthFH H E SU P C EG E b E p S  @ H C @ G s b C @ Y P @ p @ W C p C Y @ G W C U S b p G E @ A S E A W @ e s E @ P @1 H p b b cV$tVcarBd@cqhqiV$t$wqdPyQtgQ`tQqQaV$tVBt$tdWVVXpV@dV$B7V$tiQCQqVcBqQtdHtdPH @ G b @Y P @ p @ @ C A1 H p b b H U S bGE@ A SE A G P CGE S A b e P C s bG @E@G b @ G U S @ P  q$dQ`DQqdQ$aVDDtdHtdPgVQX@$cBVctRV$cVIdHtVjqdP$$Q$BaRPq$gViq ymu ~ }{ z x v

W A H C @ E @G P p S Y C @ b SE A p @ W CU b b U b @ SE A p @ pEG b G P @ H @G CG b G U Vd@DiVq`0Badth`Iqdeyo$DtVadVhdTXG|'BdGVo$DtVqcadRQIQaF$aTdt$m 87v6ih$S0VSIs d5" w 5 @ H E U V@qv4VF$$BBqhqqhaVQX@FwVQtaD#7 ath7aQcTadtidHtVw #3odCQqjqhcVs b p w E SU A C G S E @ 1 P S P @ P S G C W G C @ C G p H @ P S A @ E @ W p A b e P C 2 W@ e A pSE b H @G G P bE C b1G b E C @ W @ bG1 @G @E Y b %0 G P @ H @G CG s b W SU @ G E @ b P S 2 $dG0Qa$aTdVdCcFFddPdUt$TFQqdPbidH$`aBc`$dte a1BdQiQaF$aiqdPoXSdWht$TBqde$Dhgj) ymu ~ }{ z x v P b G C H E SU b b G b G P @G b P S hXS$FIVqPdt$$XG00RQaa$th`Y Y P @ @ W b S A W C U S P bG bE Y @ e W C bG C H @ G C H C @ b SE A p G S P U e P p S U 7 Q@`tQqpQaX@DX$chtdWV#VihXS$tdAc`$QqVBdY$FIQq$F0qe$tRBVt#|7 tDqaj7 ! H S U W Y @ e @ b HE@G@ e @ P CY p @Y P @ p @ @ C A1 H p b b H C b G C G P S P b b U h$EX$G`QcXEbtfeQtdPIcBaBq7"7 VB7V`tQqQawaVttdHtdPI&0|7 "Ft$60otd0XGqTt P bG C HE SU b b G b G P @G b P S Y @ Y P @ p @ W b S A W C U S P bG b E Y @ #hXS$FIcVqdPdD$$XG0TdQaad$Dh`7 Q`DQqdQ$X@X$$h77dWV'V&hXS$tXA`$Qqe WVBdY$FCIQq$FH&C@qedyoctBVqrw7 gtDq$r7 ! h$X$`Q$dEt6QtdPIcBaBq7'7 Vr C bG H@ GC b SE A p G S P U e P p S U H SEU WG Y @ b e e @ b HE @G @ e @ P C 2 P CY p S C P b @E A @ P C W A b C @ b SE 5 U S HE@G b @E A VB6hVTVhXS$cQ$ty`Vd@DidH$Vwqdeyo$rr7 qVcBadP"7 ! cQ$ty'0Fx

" " x $ " ( 'r &u $ x #r ! x 7 w %j x 7 ePC tV"7 ! @ Y P @ p @ G U S H E S U P C E G E b E p S  @ b G 1 @ G @ E Y b e @ G b @ E @ `tQqQ a@ q$VcV$tVcaBd@cqhq&idH$`aB$`$tq$wd $ $Bq0 $ a Qtad$Fc @ bGC @ e p b P s C H @ P S A @ E Y P @ p @ EU b H @G @ b G1 @G @ E Y b e C E @ b P S 2 b dYt0 $ qD$XGthFIath7aQ$`DQqdQc6$XG0 QaayaidH$`aB$`$tBqdecthg ymu ~ }{ z x v

UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN Department of Electrical and Computer Engineering ECE 551 Digital Signal Processing II Spring 2006

Problem Set 3
Issued: Thursday, February 2, 2005 Problem 3.1 Prove the formulas (discussed in class) for conversion between FIR lattice structure and direct form. Problem 3.2 The direct form ow graph for a causal, LTI system is shown below: Due: Thursday, February 9, 2005

z -1 x[n]
-1.5

z -1
-1

z -1
0.5

y[n]
(a) Draw the lattice form ow graph for an equivalent system, i.e. for a system which has the same impulse response. (b) Verify that the two systems have the same impulse response by tracing an impulse input through all of the paths in the ow graphs and summing the impulses that arrive at the output at the same delay.

Problem 3.3 The lattice network shown in (A) below has two inputs x1 [n]) and x2 [n] and two outputs y1 [n] and y2 [n]. If the output y1 [n] in (A) is connected to the input x2 [n], we obtain the network in (B).

Problem Set 3

x 1[n]

K K z1

y1 [n]

x 1[n]

y1 [n] K K z1

y2 [n]

x 2[n]

y2 [n]

x 2[n]

(A)

(B)

(a) Write the two dierence equations relating the two inputs to the two outputs for the network in (A). (b) Determine the system transfer function H11 (z) = Y1 (z)/X1 (z) for the network in (B). For what values of k will the system be stable? (c) Determine the system transfer function H21 (z) = Y2 (z)/X1 (z) for the network in (B). Plot |H21 (ej )| for || < .

Problem 3.4 Show that an even-length, symmetric linear-phase FIR lter cannot be a true high-pass lter; that is, show that H(ej ) = 0 for all choices of coecients. For all four types of linear-phase FIR lters (even and odd length, symmetric and anti-symmetric), determine whether or not each type can be a true lowpass or highpass lter. Problem 3.5 Approximation accuracy for FIR lter design is often measured by the energy of the approximation error H() Hd (). Show that the window that minimizes this energy criterion is the rectangular window. Problem 3.6 Dierentiator lter design: (a) Design a length-7 dierentiator lter, which has desired generalized amplitude response Hd () = j , <

using the window design method using a Hamming window. (Be sure to incorporate linear phase in your design.) Is the resulting design desirable? Do you expect substantially dierent results using frequency sampling design? (b) Design a length-8 dierentiator lter using the frequency sampling design method with equally spaced samples k = 2k/8, using whichever symmetry seems most appropriate.

Problem 3.1

Problem 3.2

Problem 3.3

Problem 3.4

Problem 3.5

Problem 3.6

x s b e P p S E E @GU C H @G C W C P C @ b G H @G s b W p @ E @ G W p S x SG e @ e P p S E @ W p S dqdPttDh${BaFQaaya$VXAdWVV7VdWdW$$QaayaTqP$XGt$Q$gq$deth{FaQqtthcq7vdethg x x e P C s SG e @ e P p SE @ W p S u s b p G P @G G @E C @ P @ G SG @ W C G Y C @ @ G s b e P p SE yVFF&tV0wyavQttth$dethgxwymV@V$dQamaQ$FQqq$aQqpVxm`Vy`gq$qPttth$ e@ bG P C p @ G b Hp e P Cpp pG P bY @ SY @ Gp ACE s S @ G U S P bGCG P@ H W A b W C bGY CE A C P QBX$RVDdq7dhXsIFtvtVxxxVB$RQd@BBd`q$oFcVoDqq$VhXS$F$RQIQX@DidHRVBdY$`Vct& dh US P bGC WC@E C b x b  b ACEs S @ G GC G Y p p e PC p p G P bY @ SY @ G @ b HE@G@ 9 C VfhXS$FQXdbVQ$00FqXs{fP6DFVtq$Ft$fct$DftV7x$RQd@BBd`q$tdPIcBaBRh x @ E p b  b P S b H @G @ G U S P b G C G P @ H W A b P C E SU A C E s S F0V$thXs{dP0ot$TdQa$aq$"VhdS$F$RQiQd@DidH6V0VjDFcV(t% hu xwysxwys x qdh u r P bG Y P pU H @G b H @G C W C P C E @ b P S 2 } |z y w u hXS$`tDQaa$$XG0Qaaya$VXAdWVfVBqe$thggymt "F 7 @ b HE@G@ 9 tdPIcBaB0qu C GE C A H SEU H @G @ C A1 H p b b H @ G E SU P bG Y P pU H @G @ G Rh$Fh$Qaaya$VtDtdHtPIq$TVhXS$`tDQa$aq$db @E @ p G C G Y p H @G C A W C e @ W C W C @E @ G @ b HE @G @ 9 dh'i$Bq0dhq'idhi('hgFt$it$rdhiQa$a'cVD1dWVIQtdpVxX1VQ$Tq$TtdPIcBaB0 W @ C S G C G S  dWQgTVxh`Y B $ Ft$oqy 'hB#0r G @ ! p S {y moh`Y B$c'I V@tVqt$dGthFIath7aQ$&`tQq b p @ e p b P s C H @ P S A @E Y P @ p 1 @ EU b SG b e S A @ E S H @G @ C A H p b b H @ G U S @ P S A @ E @ W p A b @ G e b  `$vdt$GfastdPttPh7aQ$cEV`Y 7 't Qa$aaVtDtdHtPIq$Vi$th7aQ$adtidHt$tdP{0Fx E S U p S { VDoh`Y  B $ p b U bG C @ e b P s H @ S A @E Y P p @EU b H @G  ! @ bG1 @G @E Y b e @ G E @ b P S 2 } |z y w u qstdPad$Fcvqtp$XGthFCvatPh7aQ$`tQ@qQ$$XG0QaayagymiH$`aB$`$dtq$Bqde$Dhgg ymt r {Fvs @ s C E @ G SU U S W @ A p E e P p @ C A p @ e p b P s C H @ G U S Y G @ e W @ W1 W pU C @ C 4 @ VqPVcvq$0EVg$co'V&QCdQtefqhS$VsDVCtVaVtDqVqD$XGthFIt$"V6c$BVa&QX@Q7FdCdXdWthiVFf0 H@G @ G US W@ e A p SEs @ G e Qaayaq$mVQdCQqjth$Vvq$gh @ @ ) ) @ G b W CG b H @G @ G G S P E S E @ G @ Y atPQagvq$dPhd@DF$adQaa$vq$TVq0V0Bq$Bq0oh W C p CY b H @G @ G G S P E S E@ G@ V$tVBdQaa$vq$TVq0V0Bq$Bq0g7 ccFc $ I0Vath7aQ$adtDIdHq$Rh r x E SU @ P S A @E @ W p A b @ G C @ b HE @G @ e p H @G @ G U S Y C @ E S tdPIcBaBq a dIDQIQaayaiq${VjVQVt a @ P S A @ E Y P @ p @ EU b H @G  ! @ G U S @ P S A @ E @ W p A b @ G @ E @ $th7aQ$`tQqQ$I$XG0Qaa$m{0q$VaDh$Q$adtDIdH0q$rdb 7 q $Bt0Rp 7 7 7 h G @ ! @ E @ @ W @ s e P C r E SU p e P C E SU p B#V$BqBadQ(tVF VFy dV7$cVItV VF$dF$c @ P S A @E Y P @ p @EU b H @G G P bE C b1G b p E C @ W @ bG1 @G @E Y b e @ G E @ b P S 2 Q$th7aQ$`tQqQ$$XGIQa$aIdVdCcFFddPdUt$QFQqdPb6idH$`aBc`$dtgq$IBqe$thg ymt } |z y w u

uFsVsFrpQx B@qtpVQCtccqTD"j r e Ep  | ussr p w @  pC eE p  | FVFBqtVot$ctdTTooB  | } |z y w u d7~ y{Dxvt u FsVsFriqdPctytqBom70j#hgdD"yRvg s b E A  p n n l k i e f e x w u 2 s bE@@ b s P E@G p A H S 2 e PC WC bEGY W US G P@ HGEC A@ qdPcBBtdPhqrBaqDihgfDV#VBdYca`QX@'VTRQI$$FDB9 8 % 5 4 % 3 21 % % )  %   ! !      $07600('&$#" 

@YE p S @ b S P S e P p SE W s b C US E@ S A @ b S P @ G V`cqhavahqFqDth$X@hqdPcImVB7adhqvq$Dp  U E @ G E @ P C p S @ E A E S G @ P @ G U S Y C @ E S U E @ S A @ b S P G p A G p S @ G @ b H E @ G @ 9 Y VSIHBadPbTBgatVTEthVcQ$ty'rBy$VgBqvt$VVQVBgo7adhqqDatht$tdPIcBaBoh E@ S A @ b S P G p AG p QBgo7adhqTtDaqhS @ C G @ C G b @ H @G@ e pE@ S A @ b S P G p AG p S @ G s bG p A H SY W b W A @ G p S b iHV$@q$&VQD6SgYdt0iqPbIcEBaBq{QBgo7iadhqqDaqhIt$jqdP$qih`fX$XGBYdbDy`vqhq$XG {FqetDh$EdY$B@it$dGbcFC aGqte&B@S7adhqtDaqhTV$VafiV$fq$6VQt&y$VBqjq$Vg S P p S bG H E S @ p E A @ b S P G p AG p S W CG SG @ H C @ G @ C E S G@ P @ G U S S  r b  b H @G @ G U S Y C @ E SU W @ e S H @ b S P E C @ W CE 9 C rqdsmfdPTIQaayavq${V6cVQTVdQqyivahqFQqdPdbQcRh e @ HE SUE @ A @E C P b b e e C @E SU @ b ) SG e @ e P p SE @E C G Y p e SE A W C G C G S W C @ H p #QicV$B7v$FthXS$dGttVv$VB7$XGDgox h6aQttth$vcF$`tty$t7dWV0FD$IadVviD$$% P bG CG p A H S Y @ G U S W C b bG @ H bE C G b S A1 e @ b 1 ) s b p e @G P @ H W A b W C @E C @E p s @ G BthXS$F$qih`Tq$VdWVidPdY$Bit$XGFmddPh7yQyqXGDox hiqdPctQaRQiQd@DidHqdWVTcF'Q$qhtq$Pb H @G @ G G C G @ H p % P bG Y P pU H @G @ H C @ G @ C W C r @E p b  b E S G @ P @  } |z y w u Qaayaq$'Ft$it$cmhXS$`tth'IQa$aiVcTq$VQtqWV{$qhdsmdPr$VgBttTymt H @G C W C P C @ @ GE C A b E S G @ P @ G W p S p e P p G P b Y @ S Y e @ bG P C p b U Qaaya$VDdAdWVjVv7g$FDPcVBqq$deDhtfDVC Q$RQX@BBd`jQBd$RVtj$XG& G P@ H W@ W@ e @@E G @ p W P B$RQIQX@QodCQqBcq$at&XthS ethq$$VgBqqdP$dGt$Q$iq0BthXS$`Q$6gIq$G QBB7Q$Ft$vdQiQX@QgodCQqqhg$dG0oh W p S E S G@ P s b W p@E @  P bGY@ S G @ P@@ G@ e@E C G P@ H W@ W@ e @ P S b Y G E C A b e @ b C G S H E SU @ G U S P b G Y @ S G s b p P b G C W C @ E @ e C Y C C U S A C E s S @ G C E cFDdPiQqdPV$DVcVt$V'DhXS$`QagqdPctihXS$FQXdbVQcvqVB$VBY 7VDFcVvotq$vQc9 x x dh r r P bG Y P pU H @G b H @G C W C @ G E @ b P S 2 @ hXS$`tthQa$a$dG0Qaaya$VDdAdWVq$Bqdecthgf P bGY@ P@@ G@ G P@ H W@ W@ e C s bE C e@ W A H SYY C b b  G P@ H W@ W@ e BDhXS$`QafQBgB7TdQiQX@Q&QCQqIiqdPcFD$&Qq$dbDih`BVTgtTrB$dQiQX@QodCQqox h s b b p @E W P S b Y GE C A b e @ b HE @G @ e HE SU @ G b P bG Y @ C W C @ p SG W b S A b qdPXEtQ$jXthX@dWtohicFDfPQtdPIcBaBqcVq$dPthXS$`Q$$VXAdWViatad@DX$$h7dXGb p@ eCYCY b P bGY@ C WC ES  H@G C WC US @ eCYCY C G P@ H W A b SG EC@Y@ P b VqVB$VB6dPDhXS$`Qa0cVDXAdWVVq0QIQaaya0$VXAdWVqVqVB$VB0RQIQX@DidHa$F$cQ`Qq0XGb pBIQaayaBtcVaBqhXsDVQBVBgoB$RQIQX@QQdCQtgQ$XEDdQ$XGFt$00Rhi$FDfdPghXS$F$G H @G E @ eE S1E @ b E SU p E @ @ S 3 G P @ H W @ W @ e S G @ b p @E b G C G b C GE C A b P bG C 1yQIQX@DidHIt$6$dG0Q$FDIh`ohI$FDgdPghXS$F$dQiQX@DIdHIq$gViVF$RVFqV$dDf$FIdHtqT P @ H W A b @ G b e @E C A H S Y Y GE C A b P bG CG P @ H W A b @ G U S @ s CG P C e C b e E C bE A @ H @G C W C P C @ Y GE C A U S E S G @ P @ G W p S p G P b Y @ S Y e @ bG P C p b e Qa$acVDXAdWVjVv7gohcFD{V&cVBqq$deDhtQ$RQX@BBd`jQBd$RVtj$XG&h x PC G E@ GS G PCG P SY C FVt$0Bq$V0RV$aDh`I& PhXS$FBYdbDXA$XGDqhvXDhqD{$dQiQX@QQdCQqvg"Q$XEDdQ$mFt$$VgBqVDFcVtt$Qcoh bG C W b W p H @ P S W P S G p G P@ H W@ W@ e S G @ b p @E G C G E S G@ P C U S A CE s S @ G CE 9 Y x 7 x Dxwy 7 u s C e @ @E A @ @ P C Y P bG Y P p VjQ$$Q$ty`fVBdhhXS$`tthU H@G b H@G C WC @ G US G p AG p S e PC G p A b @ s bG W@E P bGC p @ @Y P@E@ b e @ Qaaya$dG0Qaa$$VXAdWVgq$vV&qDaqhtVtDqdPgq$G qdP$FdCQ$hXS$FDdQ`tQ$B7tqT x @E p b Fi$qhXs{
b x[n] d z -1 c y[n]

b 0 x[n] z a (a) -1 y[n] x[n]

b 0 z b 1 b 0 -1 y[n]

b 1

a (b)

x[n]

-1

z b 1 a

-1

y[n]

(c)

r @E p b v$qhXs{

UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN Department of Electrical and Computer Engineering ECE 551 Digital Signal Processing II Spring 2006

Problem Set 4
Issued: Thursday, February 9, 2005 Problem 4.1 Design a length-2 least-squares optimal anti-symmetric FIR lter for a highpass lter with s = /3, p = /2. The weight function is 0 in the transition band, 1 in the passband, and 2 in the stopband. Sketch the frequency response of your lter. Problem 4.2 In this problem we shall design a high-pass FIR lter by minimising the weighted L2 error, i.e., h [n] = argminh[n] 2 = The specications are given below: (1) M = 10 (2) Hd (ej ) = ejM/2 , 0,
2 2

Due: Thursday, February 16, 2005

1 2

W ()|Hd (ej ) H(ej )|2 d

< || else

10, 1, (3) W () = 0,

|| < 3 < || else

(a) Solve the integrals (refer Lecture 16 notes) to get d[n] and Mm,n . (b) Use Matlab backslash operator to nd h [n]. (c) Plot H(ej ). (d) Verify using the command rls.

Problem Set 4

Problem 4.3 In this problem, you will design a length three FIR lter to approximate the following frequency response: Hd (ej ) = j, || < by minimizing the following error criterion, E= for W () = 1 2

W ()|Hd (ej ) H(ej )|2 d, 1 , 0 < || | sin()| 2

(a) Let h[n] = a[n + 1] + b[n] + c[n 1] be your length three lter, where a, b, and c are real constants. Show that in order to minimize E above, that b = 0 and c = a. Note that H(ej ) is the discrete-time Fourier transform of h[n]. (b) Find the lter of the form h[n] = a[n + 1] a[n 1] that minimizes E.

Problem 4.4 Design a length-3, symmetric equiripple high-pass lter with the following Ad (ej ) with a stopband edge of and a passband edge of , and with uniform weighting of the stopband and passband. 3 2 What are the lter coecients and the ripple amplitude ? Ad (ej ) = 1, 2 = 0, otherwise. (1) (2)

Hint: You do not need Parks-McCellan algorithm to solve this problem. Problem 4.5 Problem 7.36a-f in Oppenheim and Schafer (2nd Ed.) Problem 4.6 Problem 7.38 in Oppenheim and Schafer (2nd Ed.)

Problem 4.1

Problem 4.2

Problem 4.3 a)

b)

Problem 4.4

Problem 4.5

Problem 4.6

x pB$tDqdP&BBX$RVDdft$$XG0vQaFdCQ$$V`tDIdFD$adhqfadGt0VX$XGDtVQaqDXca$dtXcVdUttp G p A b E@ bG PC p @ G b e@G W@EESY P p b GC G @ b S P @ b @ b b e eC e@G p bEG b e W HES b P C P bG C W b W p H @ @ G H SEU s b W p @E E SEE @ P bG C bG P C p @ G s W @ e S 4 P bG C W b W p H W VthXS$FBdYbDXA$XGtiaQq$h$IqdP$XGD$Q$cV$$BhXS$FQX$dVtt$qdPbQqf0BthdS$FBdYdbDXA$dGt"dWVC E SU e @ p b s b e P p S E b b G @ H b E C b 1 VQ$tdtdPttthcg$XG0dY$Bit$dGcFvXGDox hah7tt$0I7Fm dPQqdPIBaBqhV @ S A A p S cE DA b e @ b H E @ G @ e p S G C @s PCE @ G E@S e@G p bEG b e W HES b P p p@Y SE A H S e PCE @ b C @ S VqVct$BVoQaqDdcaadt&XIcVXUtDB$Q`y$thqtVvaXGt07aG G p A b @ G E@ b PS 2 tDqdPq$TBqdecthg 7 7 77 { S E@S b voD$BVodWdW0 E S TVqP cBq0 E@ b@P WC ES Bq$XGQqp dWVvVU XUFt$gcD$ gVqdpVFaQV$FdCiq$qPIcBaBFx b GC G Y p US @ WC G@sE W @ G @ b HE@G@ 9 @E@ o x x dr q { HE SU @ G U S b P bG CG P @ @E A @ cVvq$mVdjhdS$F$RQaQctB$E GRQiQd@Dih`00dEQq$TFt$fct$0thXS$`VcQqhXs$ehqBq$tXpV@DVthVt$$Qd@$FQqt$dGthFIH P @ H W A H S Y r b @ G G C G Y p P bG Y CEU e @ P b W S @ G b @ P S P C G W @E C @ e p b P s C @ S E@ H p P W P S W S PCY P bGCG P@ H W A b b G b e@ p E@G b s@E ES H@ H @ G GC G @ H p % $hq0qBtiXthIehqVBIhXS$F$dQiQX@DiHrdt$dP&QatrBaadhB$i$VIQit$Ft$0it$$0r 77 " pV@$db0$BqS E @ $V G ys Fx 6 R Fr ys Fx 7 m07 T7 7 7V 7 htt7V r r e PC @ b HE@G@ jtVqdPcBaBqe

U TFx

E P bG C bG P C p U S @ Y P @ C @ G b VS hXS$FQX$dVtdV&`DQaDFq$Php r

ePC tVrp

7 #7 7 ESEE@ S e P pS V$$BgFqtthcE p @E p b b p@E@ '$qhXs{ dPqV$Bq

R7 7V

R7 "7V

W pU @ p P b @E A @ W C bG C H @ G C H s b W SU @ G e P C H p S p W SU G C G GE C A @ G P |  w thBatDhXS$$Q$D`VBY$FIQq$FIiqdPoXSdWhvq$DDQIjhVqoXSdWhTFt$T$$FDvq$ot6
z x[n]
-1

W@ P S P bGCG P@ H W A b @ G b H@G  ! WC pCY C E@ b P S 2 oXSQ70q$hXS$F$dQiQX@DIdHq$$XGQaayay#V$tVBIBqe$thg ymt } |z y w u uFsVsFrpQxg$EFCtcEDBqtVQtccqTTD"j p @ pC eE p  | ussr p rx EC pE @  pC eE p  | FVF7QcFtctBqtVot$ctdTTooB  | } |z y w u Id7~ y{Dxvt u FsVsFriqdPctytqBom70j#hgdD"yRvg s b E A  p n n l k i e f e x w u 2 s bE@@ b s P E@G p A H S 2 e PC WC bEGY W US G P@ HGEC A@ qdPcBBtdPhqrBaqDihgfDV#VBdYca`QX@'VTRQI$$FDB9 8 % 5 4 % 3 21 % % )  %   ! !      $07600('&$#" 

a w[n]

-1

y[n]

r G S P b W A @ p G b H @ G Y p S P U # # ! @ b H E @ G @ e e P BVqd0dPVdC`q B$$dy`@ QaayajcD$Iq{q '! $" qdPIcBaBtjtVC pBDhXS$XGtth`aQt$aF$6a '" 7 P b b e P SY @@ G @G CG  #!  G C G Y p e tpm U @ &(Ft$t$&t7 hS 7r P H@G C PC QaayaVBY P @ b p H @G  ! P C SG G P W C b p @ ) H @G b Qq0{V@RQa$a'yIVa'RQX@VFXtQQa$ard{0 DVC eP PS P b bePSY GC E@eP h{DhXS$XGtth`Ft0'Bqt0r G S P b W A @ p G b @ H @ G Y p S P U # ! @ b H E @ G @ e e P BVqR0PVdCDy`qB$adq`Qaayajt$Iq{r0 '" qdPIcBaBtjtVC p P b b e SY @@ G @G CG #! BDhSX$XGttPh`iaQt$aF$y a $& 0  G C G Y p e P p U @ #! H@G C P C 0&%Ft$cD$6DthS 700 $" Qa$agVBY P @ b H G  ! P C SG G P W C b p @ % H @G b e P C P S P b b e P S Y G C E @ e P Qq0mV@RpQ@aayamyiVvamdQX@VFXtQQa$a{dm tVm$ hmthXS$dGtth`'Ft0{Bqt0Fx 7 7 { W @ b G Y @ A @ E VdQVX$`Q7aQ$tp tVC eP U S H E SU P C EG E b E p S  @ b G1 @G @ E Y b e @ VcV$tVcaTBX@qhqvidH$`aBc`$dtt$G @ SP@ e C aGVqQqe"  DPV tV"$7tV"$ V{cV$tVca{BX@cqhtt$aVqQq"0tVm0 B! e P C e P C U S H E SU P C EG E b E p S  @ G @G S P @ e  e P C G@ 7 m V 7 m  m p 2 $ a{Vv9 p @ G E P S Y @E b @ G C E @ W  ! @ G @ p E @GE @ P S Y 2 9 @ G G C G e P V@btBcEBadGWDPhXSb$`YtpaGath`Q$XEQt$TVg$mBaXGDgyvq$v$t0cBacBVRh`(Vq$TFt$DVC  77 {    9 D x02 D a 7 9 xT2 V@bqWVCQ@tdeb@$FcBa$BVdh`xTq$TFD$it$$VBIQaayag$Qq$Vt p E C E@GE@ P SY 9 2 @ G G C G @ H p C p H@G S G @@ G E S ) Qaay H@G 7 m 0 amVv9 p 2

% H@G Qaay 9 p 2 a D $mVv9 D xT2 D $ e P SY@ E@ A W A HC x @GCE s W A HC GC @GCE@ AS W th`QaTB7Qd@DiV$ o aFcqdPdbDIV$0FaFcB7VWVC E @GE @ P S Y 2 9 e P C E @GE @ P S Y 9 2 @ G b b H @G S G s b W SU @ G E @ b P S 2 } |z y w u cBa$BVdh`Vv6tVBa$BVRh`xTq$dYt0P0IQaaya&iqdP0XSdWhvq$Bqdecthgymt t E @ G G @ @ G C p u @ E p b  7 S E S @ E h b  U S G C G p P b G C G P @ H W A b b u @ E p b  b P b G C G P @ BaaB7q$Vttq$thXs{mVU TVcq7Xs{mV0Ft$yhXS$F$dQiQX@DiHcdYtqq$thXs{dPjhXS$F$dQiH p 1 W A b @ G E SU Y e P C `X@idHIq$vVohtV gtVC dPot$BVatdPVhFjtdP$`QaV$tqh$cFDFQ7BT0w eP b S E @ S G b C s C s bG Y @G SE A S P GE C A G C @ A @ e PC @ b HE@G@ tjtVvtdPIcBaB9
c x[n] z v[n]
-1 d

W @ P S H E S U @ G @ C G H @ G @ G U S P b G C G P @ H W A b @ b G C P E @ W C 7 P % voXSQ7oq$Vvt$0QVF$Qa$aq$"VhdS$F$RQiQd@DidHVd$FtcBaXGVj0qu 7 o x " m b C e @ G P @ @ E A @ E @ P C Y b @ W C W C P S @ C G P C 7 DdPTV6QaRQaQctB$7jVBcYt0QqpVx7dWVfhVF$jVBY eP tVC p P b@ e hXs$Qqf s p S W C G C G @ G S @ b S P 7 { S E E @ P b G C b G P C p W C S G @ p hqhq$dGVFD$aV$dhqhi cVE $$B6hXS$FQX$dVt"WV&aqte GtDaqh0t$VrdQqh7ih`q$V`tVP dCcFF0t$'db t VthXSVC p A G p S @ G U S G P @ P S A H S Y @ G U S @ Y b E C @ G e Vdp @ P W PC S G @ p e W C P b G p A G p S @ G U a0tttVthXscrqDaqhq$VS G P@ PS A HSY @ G US @Y P bEC @ G @E@ RQqh7ih`q$'V`tVdCcFFt$0db $ $Bt07p $ x$  GC G WCY@ tvFt$"dWVBQT tdb  @E pG Y pEG b G E SU  bG CE @ b S P P bG C bG P C p 1 SG W C P b @ G G C G Y p e P C @ b HE @G @ cq$`tcaagdt$V'tjXS$Fc0$dhqhXS$FQX$dVtFa1VthXscq$Ft$&t$gDVfTqdPcBaBqe

y[n]

W@ P S C W@ e S H @ W p A e PC @C G P WC b p @ PC s p SE vXSQ70ot$0VQqyiadDDtVDBaVRQX@VFXtQjVhth$q$G e @G P @ @E A @E @ P C Y @ Y P @ p @ G p AG p S SG @ Y P @ p @ G p A b H SEU @ S C H @ W CE @ S @ G p P @ S QaRQ$Q$tB$7VBv`tQqQaqDaqhia`tQtdQaqDtdPh$Vo7FQaaG yadWV cBVq$qQahtcY XQaFdCcDV$ttFdb `FD$tVd#QaXGIdHdbDtVD&X$RQd@BY$v$Fa'WtVTa $ Ft$qdPt$$$0Fx W@G bE ASE A AC T G C G e P C p e @ b W e P C WG P b Y X p @E C R e P C p G C G s b H p % E@GE@ P SY 9 2 WC@ b PC b e W A HC P@ QBa$BVdh`xTVQqejV$XG0QX@iV$fQt$G e P C p R @ P S A @ E @ W p A b b E @ W @ b @ Y @ E @ G b e @ @ E @ W p e P C @ C S G Y C e @ G W p e S H P @ tVta'VaDh$Q$$dtDiH$XG0BaXGDVXQ`Q$q$$XGQB$BaXGDyDVDBaVDIacVDQaFdCttyijQt$G drVthds$tVa$VDq$'QBVXQ`Q$&q$$ ath7aQ$i$dtDiHVtgQqtVDcIdH$Dhqtdp$dh`q0 b WC P b e PC C A @ G pE@ b@Y@E @ G G % @ P S A@E @ W p A b C W@ P PC Y @ bG p S P bG P SY @ o cVBVQXG$tVat$`tD$DiaUVI`tQqQ$TBaRQ`iiattVDhXs$q$gQaFdCttyiQq$DVC G HE SU @ C b P C EG @ G @ Y p e SE A SG T U S Y P @ p @EU E @G P @ Y C SG A p W C P b @ G @G W p e S H P @ G e P a BaXGDqPDFt$`atD&q$qdP$Dr7 thXscidH$athqdP$Rh`ga`tQqQahXS$FIcVqdP&q$$$BVdh`Y E @ W s b A C 1 @ W p A @ G s b p VC {P b @ b G 1 p S p b G P S Y C S G @ Y P @ p @ P b G C H E S U b @ G G E @ P S W H@G @ G US E@G b H PCEG @  Qaayaq$VBaaXGI$DVcaqT P @ b s b @ Y P @ p @ P bG C HE SU b @ G @E @ p P bG W p e S (QVXh`tQqQahXS$FIcVqdPq$g$Bq0hXS$FdCDtIH @ e p W A H C1 @ W p A s b p H @G P bG C b P p H H S Y e P C C A C G b A @ e @E p s s b W SU @  } |z y w u qt$XGbDiV`adDDtdP$tQaayahXS$FBYttIh`DVDa$VD$`dYBq$qhtqdP0XSdWhtTj ymt 7 m7 m b W A @ p G S P U R Y p @ P S @ b H E @G @ e p S U dPVdC`tBVtmq$'SctcvqhqdPIBaBqDqa{j P G C G Y p R E @ W @ b @ Y @E W CG C @ S S Y SG W b S A b b p GE C A b P @ b s C b Ft$cD$a'BaXGDIVXQ`QciX@DF$ajahdtcfaX@DXc$h7XGd"7$FDgdPQVdhvV$ $XG
T

7 m 7 m C6Q$BVo`Qcb b e@E@SY@E U S P p @G C bG @ P C e P C e @GE SG b e P p e @ b @ Y @E b H W C P b @ G @ S A A p  Vqp QVaFIH$aQVtVQa$VaatttjQVdQ`Q$dg$d6VthXscvq$$hDqy P bE@ bE hXSBaXG c`Y G b p @ G U bG C @ S e p @ W p A G b e P C 1 H p b b H 7 975 C A W W C @ b @ G b U $dtq$gad$F$$ Qdq{VadDD$tde0qtVtdHtdPI'p4GG'&$VDTXSrVQtdeq$da Tr
s(t)
T, h(t)

dCyd$`Q$XEtQqPV$tV7jVBg$dIFD$ctca 7Ft$adtq$tdP{0Fx b WG Y @ b e e @ b CG S @ P C Y H G C G Y p @ A C @ W p A @ G e b  E @GE @ P S Y 2 9 @ G U S E @ W P bG Y pEG P S Y @E 7 975 @ A C @ W p A @ G b @E @ QBacBVRh`(Vvt$mVrBadGDhXS$`taath`Q$tp4G8'i7Ftc$dtDvt$d$ $Bq


x[n] T, g(t) s(t)

CQaaXG$tVcavcFDp b e@G b H PCEG @EC


z -1

@E $FC

7 { 7 { dWVVFx Fx DB W C E SU E C dpA@86V$FtdP 7 975 p E C b dh7ya#VDdPhXscVq$v$BqQaaya4$th7aQ$#VC$$FD2qdPoXSdWht$Bqde$thg0ymt W S H W C b bE S @ G @E @ H @G 3 @ P S A @E W bGE C A1 s b W SU @ G E @ b P S 2 ) } |z y w u


x[n] + y[n]

y[n]

f(t)

D/C

D/C

C/D

s(t)

^ x[n]

p WS H @P @ yBhya Bqvq0

7 # P b b e PSY @@ G E@ e P BthdS$XGtth`$Qq$Bqttp '! qdPcBaBqIad% @ b HE@G@ e S W { 7 m '! g # P g 7 m 7 m @E @ b p H @G  ! P C @ W @ C @ $Bq0DXpV@DQaayayjVvVXdbVQtQ7 ftDaqhS P G p A G p S aG G p A H SE H @G W CE @ S @ G b @ A C @ W p A @ G P S P S b e P S Y G C E @ e P qDqdPb hcU QaayaTWVcBVoq$dWdWam7Ft$adDDq$hthaXGdbDth`iFt0iBqt edbVxId0QqyiadDDIdQX@VxdtdQ6q$tVQDad$F$fcFff$FA h$ithXS$dGtth`fq$fit$$0 W C b W @ e S H @ W p A G P W C b p @ @ G e P C p e @ bG C @E C x GE C H SEU P b b e P S Y @ G @ H p % T Y P @ p @ E U E @ G P @ Y @ G e P C p d e P C p W `tQqQ$BaRQ`q$tVWDVD0 p b p HE SU P CEG E bE p S  b @ G E S p R e P C p W C P b b C H S e1 @ bG @ G E @ b @ U S HE @ 0IQIcVatVcamBX@cqhtmXEQt$"VV$'DV#$ $ dVthXs$dPVIht`idH$gq$"Bt$XGQtV"IcBaG PBgatVqhV$Q$ty`QH hqjamVqdPIcBaBtqF$FDdPQtdBQ$thXS$dGtth`t$Bqt0r b E @ P C E p S @E A @ C p S  @ b HE @G @ e p x GE C A b e @ b Y @ A P b b e P S Y @ G E @ e P @ H S E SU W S $m(ih$0VDdehq W@ e S H G P WC b p @ b G GC G Y p QtiRQX@VFXtQdD$TFt$jt$ idH$qdPdbiV$q$jtVC eVTIcBaP0WtV0 @ bG s W A HC @ G e P T US HE@G b p d e PC p p b G b e P C @ G e P T Y P @ p @EU E @G P @ Y @ P S P b b e P S Y @ b Y @E A @ G @ b HE @G @ 0Ict$qdeqtVD6q$vDVC a`tQqQ$BaRQ`gq$G hIthXS$dGtth`6$dBQ$tq$ftdPIcBaB9
T x[n]
x[n] s(t) Tansmitter

Samples to impulses
Channel samples to impulses

h(t)

r(t)

xs(t)

shaping filter

g(t)

equivalent pulse

p(t)

r(t)

-jt

jt

Receiver T

receive filter

Re{ }

f(t)

T C/D
C/D s(t) x[n]
^

x [n]

UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN Department of Electrical and Computer Engineering ECE 551 Digital Signal Processing II Spring 2006

Problem Set 5
Issued: Tuesday, February 23, 2006 Due: Thursday, March 2, 2006

Reading: Lecture notes 13-18, Chp. 7 (Oppenheim et al.), and Chp. 8 ( Proakis et al.) Problem 5.1 In this problem, you will use Pronys method (sometimes called the Extended Pronys method) to nd a rst-order approximation to the IIR lter with impulse response hd [n] and system function Hd (z) = 1 (1
1 1 2 z )(1 1 4 z 1 )

That is, you will determine the coecients b0 , b1 , and a1 in H(z) = b0 + b1 z 1 Hd (z), 1 + a1 z 1

in a quasi-L2 method. Note that Hd (z) satises the following dierence equation:
N 1

hd [n] +
k=1

ak hd [n k] =

bn , 0 n M 1 , 0, else

where N = 3, and M = 2 for this second-order lter. Pronys method uses this relationship between the impulse response and the lter coecients to nd the approximating lter by setting N and M to the denominator and numerator orders of H(z), the approximating lter, and then solving for ak and bk in a least-squares fashion. (a) First, determine hd [n] and set b0 = hd [0]. (b) The second step is to solve for the coecients ak which minimize the expression
N 1

|hd [n] +
n=M k=1

ak hd [n k]|2 .

Verify that the minimizing ak s satisfy:


N 1

ak
k=1 n=M

hd [n k]hd [n ] =
n=M

hd [n]hd [n ],

for = 1 . . . N 1.

Problem Set 5

In this case, you have M = 2, and N = 2, so this reduces to the single equation:

a1
n=2

hd [n 1]hd [n 1] =
n=2

hd [n]hd [n 1],

which you must solve for the coecient a1 . Use this equation to solve for a1 . (c) The last step is to solve for the remaining numerator coecients, bn . Verify that the coecients bn should satisfy:
N 1 k=1

bn = hd [n] +

ak hd [n k].

Use this expression to solve for the coecient b1 . (d) Use the Matlab function prony to verify that you have the correct coecients. Plot the impulse response of the lter you obtained and hd [n]. Is it a good match? (e) Repeat this problem, using Shanks method. Which has a better impulse response match?

Problem 5.2 Given a desired impulse response hd [n], Pronys method solves for the set of lter coecients (numerator coecients bk , k = 0, . . . , (M 1) and denominator coecients ak , k = 1, . . . , (N 1)) which match the rst M samples of the impulse response and minimize a least-squares criterion over the remaining samples of the impulse response. The criterion which is minimized is
N0 N 1

E=
n=M

|hd [n] +
k=1

ak hd [n k]|2 ,

where N0 can be nite if the approximation is desired to be based on a nite-length segment of the impulse response. We can dene an error sequence e[n] which is minimized,
N0

E=
n=M

|e[n]|2 .

where
N 1

e[n] = hd [n] +
k=1 N 1

ak hd [n k], M n N0 ak hd [n k] bn , 0 n (M 1)
k=1

= hd [n] +

Problem Set 5

(a) Let h[n] be the impulse response of the lter designed with Pronys method, i.e. the lter whose system function is given by
M 1

bk z k H(z) =
k=0 N 1

. ak z k

1+
k=1

Determine an expression for h[n], n 0 in terms of the lter coecients ak , k = 1, . . . , N 1 and bl , l = 0, . . . , M 1. (b) Let [n] = hd [n] h[n], n 0 be the impulse response approximation error. Show that e[n] and [n] satisfy
N

[n] = e[n]
k=1

ak [n k]

Problem 5.3 Consider the discrete-time LTI system with input x[n] and output y[n] satisfying y[n] = x[n] + x[n 1] + x[n 2] + x[n 3] + x[n 4] + x[n 5] (a) Determine the impulse response h[n] for this system. Now consider approximating this system with a second-order IIR system with system function H(z), i.e. 1 H(z) = 1 + a z 2 1 + a1 z 2 using Pronys Method. To do this, determine the coecients a1 and a2 to minimize the following criterion:

E=
n=

|h[n] + a1 h[n 1] + a2 h[n 2]|2

(b) Write a set of two equations in the two unknowns a1 and a2 and solve them for a1 and a2 . (c) Determine the approximating impulse response h[n]. (d) Determine the value of E for this solution and comment on the quality of this approximation.

Problem 5.1

Problem 5.2

Problem 5.3

p b E @ S A @ b S P W CG SG @ G @ b HE @G @ e p S W % b x b @ b S h7 Q''o#7 dP0Bgo7adhtV$Vavq$qPIcBaBqDq$d RdP$ FQdv$dhqP P bG C bG P C p @ G SG @ p e G p AG p S @ G U S H pEG Y @ A E @ S A @ G G C G S  W @ bG Y @ A @ hXS$FQX$dVtdq$ atttDaqhq$qVT a $gtca`Q7aBgo7t$Ft$otyVdQVX$`Q7aQ$E p7 RjtV#7 miaqD7 htV#7 qV$dQqh7ih`g"VIhXS$dG$h7$B7q$q$d#7 qaqhqT0r e P C SG @ p e e P C G P@ P S A H SY S G U S P b b S AE@ A p @ G b G p AG p S @  e P C p p s bG W @E P bG C p @ @Y P @E@ b 7 RfDVt7 mt7 qdP$FdCQ$jhXS$FtQ`tQ$B7te @ G @ b H E @G @ e p S W % SG H S EU e P C SG H S EU P b G Y P pU E @U P C EG @ G @ b H E @G @ 9 q$qdPIcBaBtDqadr7 Ia7 Rh$jtV"7 Ia#7 mh$hdS$`tthTB$tVcavq$tdPIcBaB0Fx

x a t C e @ P @ e b bG CE @ b S P1 P bG C bG P C p 1 SG W C P b G p AG p S @  W @ bGY @ A @E p e P VQqDBqTIXS$FcadhtyhXS$FQX$dVtFa1VthXs$TtDaqhqTVXQVX$`Q7aQ$q7 dftVC 7 ma0qtVaFIdHBQtq$VgqDath0q$aVqQq7 VtV7 tFBq BDVthXs$$XG0tVC SG @ p e ESGC bY@ e @ G US G p AG p S @ G @GS P@ e e PC G@ ! E@ S A WC P b b e P s '7$cqBg p WCY bY@ A p H pEGY@ A VXdWVBDdBQ$yjtca`Q7a b P@ e E@ S A e@ b W e P C C b W C P b H S e P CE   P C@ H1 SE@ C b G C G @ H p % dG$tQqfBgo7QaXGIHdbttVD$XG0VthXschqtVcDVQiF$BBfd7 mjFD$it$$0

@ b A@G P bG C bG P C p @ G SG e P S A @EE SY b e @G W @EE SY P p e P P @@ G@ @ e p W A H VBX$fBaafhXS$FQX$dVtdq$Iatth7aQc$V`7 {$XG0QaFdCQc$V`tttVC gQBB7tt$XGdbDIVC PQaqDXaadtXIVXUDtrV`tQqQa&adhq&aXGt0VQIF$BBd7 RIFt$&itc$VQX@DV$tdD$G b e@G p bEG b e W HE S b P p p @Y P@ p @ @ b S P @ b P C@ H1 SE@ C b G C G @ H p C @ H W SE A b P P bG C bG P C p SG @ p e E SEE@ @ G @G S P@ e @E@ G p AG p S E@ bG P C p @ (#hXS$FQX$dVtdajqD&V$cBfq$&QaVqQq7 Rg$Bq07 R"7 m7 qDathBBX$dVt6q0 ys q$cVI  x GFD$GiaVSaFIHBYQ@qvq$GisqdPbDQ`Q$tBaXGDqP$VdCdbV1$dV#VQqejVddhh C S E GC b e @ e @ Y @E A E @ W s b W C bG P C W C @ b P C b qTx qhF etVC qx rhQ P x b w b  b P S H@G @ G E@ b P S $XG0dxqXs{dPj0q$Qaayat$Bqde$th2 dxcqhXs{ w @E p b
d1[n] H1(z) Q H3(z) M

s b A C @ b S P E @ e E S1 e P S Y @ E SU H @G @ G @ W C P C @ qtdPDFt$ahqBqVyth`Qa0VQaayat$vBddVtVg H W SE A b G P s b A C 1@ b S P E@ eE S1GE b s W A H CE@ S e @E@ b P SY @ @E pGY W P Qd@DV$tt$qqqdPDFD$`adhqBqcVat$XG0ItdPdbDiVccBVoQcBqde$th`vcq$`QX@f ymt } |z y w u uFsVsFrpVr$EFCtcEDBqtVQtccqTTD"j p @ pC eE p  | ussr p x EC pE @  pC eE p  | FVF7QcFtctBqtVot$ctdTTooB  | } |z y w u d7~ y{Dxvt u FsVsFriqdPctytqBom70j#hgdD"yRvg s b E A  p n n l k i e f e x w u 2 s bE@@ b s P E@G p A H S 2 e PC WC bEGY W US G P@ HGEC A@ qdPcBBtdPhqrBaqDihgfDV#VBdYca`QX@'VTRQI$$FDB9 8 % 5 4 % 3 21 % % )  %   ! !      $07600('&$#" 

x[n] +

r[n] +

u[n] d 2 [n] H2(z) -

y[n]

w[n]

v[n]=w[Mn]

r w u r x j { yFgj { y j { yj { y j { y7 m7 s b U b G C G p AG p S e P C G p A b b H @G  ! @ b G1 @G @ E Y b e @ G E @ b P S 2  } |z y w u qdPy$d$F$7 qDaqhtVm7 mgqDqdP$dG0QaayayvidH$`aB$`cdtq$Bqe$thggymt s Pb A C @ b S P U S @ S A E p A @ G b G C qdDFD$vadhq'Vah7$qDq$d0FtTqu @ b S P P b G C b G P C  d @ G S G @ p e G p A G p S @ G U H p E G Y @ A E @ S A @ G p @ b H E @ G @ e e P VadhqjhdS$FQX$RVD t$IaqtqDathvq${VS tca`Q7a0Bgo7vt$t a gqdPIcBaBtjtVC p p e P C  E SU e P C E SU P b @ E A @ s b W p @ Tdh tVp  dh h V0dhRBigtV0dhd IV0thXS$cQ$tyIqdP$XGD$Q$E @ G @GC WC@ P@  s q$iaFtdpVFBQqTy0dh IVFjaXpV@7 oFV`Q`q$aFtdPdHdbQadhViahq @ C H SG b p U S G Y @ @ @ G @G C b W @ SG @ S S 2 r1 w @E p b r$qhXs{
x[n] +
xc(t)

e P C p p U S HE @G b p dhVitVdh&dhVIDdho'VcBadPh t7 7Ia"7 Foh$hXS$`tthU p SG H SEU P bG Y P p E @U P C EG @ G e P C p p SG BaDVcaq$ftVdhd I7 am7 h$fhXS$`tth0BatVaq$7dhRBi77 a"7 {h$U H S EU P b G Y P pU E @U P C EG @ G p p SG H S E P bG Y P pU E @U P CEG @ G e P p e W @ W E @ bG P C p @ G E SU W @ e S H @ b S P @ b b e e C @ G s b hdS$`tthBatVcaq$t"RcQX@Q7FdCh&BBX$RVDd&q$VqQqyi$dhqVX$dGttVt$jqdP$ VFDTyqBah Qaaya y6V7QqyiiqdPoXSdWhvq$Btde$thgcVDqQB0BqVg$ct&XthS @ s C A G @ P @ @ H @G 3  % 4 C E SU W @ e S H s b W SU @ G E @ b P S 2 Y C e @ @U E @ e E S G E W P b s b A C @ b S P E @ e E S1 b @ b Y C SG e @ P b @ e @E C p 3  % 4 C SG e @EE @U @E p A S W s b A C $dG0qdPDFtcIadhqvBqVyhXst&VBX@tcV6agQqhXs$Qq$Fy6V6agQ$$BB$#BDVXSqdPFt$ @$dhqVF$a1$XGDfBQd@$XGdbdWDF$$tdPb h$mB7y$XdWVBdYDmDVXS$VDqQB0tdPDFt$`$dhq'BqcV{BqhXs0r b S P @ s CG b W p 4 b b CG H SEU E @ p W C b A G A S W Y C e @ @U s b A C 1 @ b S P E @ eE S E @ b 3 @ bS P P bGC bG PC p s b W p@E @ G e PC E@ W V$dhqjhXS$FQd$RVtqdP$XGtcQ$vq$DVTBaXGD sqP$VdCdbV1$dVq$q$`Q7xqdPdbDIV$cBVoqhmQ`tQadQaB$q$'VgdPvXWVBDXBQ7a'RQIIhg b W C bG P C @ G GY@ C s W A H CE@ S S P S @Y P@G P@ @@E G E S S G b W CY bY@ A G P@ H H S 2 E@GE@ P S 9 % s W A HCE@S PC b x Ba$BVRh`Y xqdPbDiV$cBV(VdPo XpV@hqdPdbDIV$cBVoVgah7$qDq$dfFtTFx b s W A H CE @ S U S @ S AE p A @ G b G C

U S G P @ e P @ A@ e 7 {"VdQqtQ7BqtPb e P C r e P C r P@@ G@ e@G p bEG b e W HE S b P p p@Y SE A @ b S P @ b E C P bG CG @ P@1@ b C tVvFvtVvF06QBgB7QaqDdcaadtXIcVXUtDRB$Q`y$tT$dhq0aXGD0$FthXS$F$aT$tQa`qdevrb @E @ p p @ S C e W @ W E @ bG P C p @ G E SU W @ e S H @ b S P @ b b e e C @ G @ 7 R$Bt0D7 dy7 m7 V7FcQd@QFChBBX$dVtq$0VQqyi$dhqvVX$XGDtVq$a x1 w @E p b Fcr$qhXs{
T/M C/D

E @GE @ P S Y 9 % s W A H CE @ S P C E SU W @ e S H H @G s b W SU @ G E @ b P S 2 } |z y w u oBa$BVRh`xIqdPbDiV$cBVV0VQqyiQaayaiqdPoXSdWhvq$Bqe$thgymt e P C P@@ G@ P bG W@E @ G bY@ A 6tV6QBgB7hXS$FdCQcq$XUBQ7a pQhBQaVQ$`tPb VvQ`DtQ$f7VBI$XGD&BBX$dVt6q$G Xy$dYtoqTp &B7iqDqdPtVC p @ C @E Y C e@Y p e@E @ PCY b E@ bG PC p @ W b p S E@ S A G p A b e P  P @ b s C E SU p b Y @ A r U e P C p p U S P b G Y P pU C C  @ G @ b H E @G @ tQVXhI0VDVXUBQ7ap D ' jtV'VjhXS$`tthi0Vtq$vtdPIcBaB9

x2[n]

Analog anti-aliasing filter

H1(z) + H2(z)
x[n] Q

Q1
y[n]

y1[n]
Digital LPF c=/M

Q2

y2[n]

D1(z)
w[n] M v[n]=w[nM]

D2(z)

y[n]

P bGC b SE A AC b hdS$FIdHq$tDFdt$G UVXGdbVCDdvq$GhS0RPQiHhS`tPVPhXS$GqdphS$gbt$TVU4VS$cBhXS$FtQq$"VqpVxq$tdPIcBaB0qu S W p @ P G @ H Y e C b W G ES & E EE@ P bGC p @ @ G US @ WC @ G @ b HE@G@ 9 @ P A @ @ p A b s bG C b SE A A C @ G @ b HE@G@ 9 7 m  $thS7aQ$EadWtDIdHiqdP$FIdHqcttFq$tdPIcBaB0 tVC 4VQq$G  e P  E SU H @ @ WS e C e W G @@ GC VXhatPVQ@VXhSa@7Sa&eQBqPFt$G 3etVC 2tPoqyqPtISgG@q$dPthXS$FtQiVBaaXGcTr  P  S P p G b P bG C p @ S G U S G@ C @ bE 0 1)
'

r j 

x y j "

y7 "

'

&

ESEE@ P bGC p @ s b WS oV$$BhXS$FDdQqdPoXSdWhU @ @ b q$GBXbIdHtPbIH aG %etVC $$GRQX@bB#d`q$jqdPcBaBqBdt$G qe "yq$Bf& dh$qqP$tp S P  P Y "@ SY @ G @ b HE@G@ e p b S S  e S G@ 4 ! P SE 5 s b y  y  rx qh  x b p  V@Ddh PhdS$`YtthUQa$a$GX0Q@aaya${B@qcEV1yeth`Q$$XG0Qaayamdt$0tdP$FIdHq$DtFmBtde$th`o bG P p H@G b H G E e S P SY@ C b H@G b G s bGC b SE A AC E@ b P SY S H @G b G E SU  @ P S A @ E @ W p A b @ G @ b H E @G @ 9 Qaayadt$0V7 "ath7aQ$adtidHq$tdPIcBaB0Fx

UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN Department of Electrical and Computer Engineering ECE 551 Digital Signal Processing II Spring 2006

Problem Set 6
Issued: Thursday, March 2, 2006 Reading: Lecture notes 25-27, Chapter 11 ( Proakis et al.) Problem 6.1 Let x[n] =
N n k=1 ck k .

Due: Thursday, March 9, 2006

(a) Show that the sequence {x[n]} is N th-order linearly predictable. (b) Develop an algorithm to determine the ck s and k s from {x[n]}L1 for L = 2N . n=0

Problem 6.2 The power density spectrum of an AR process {x(n)} is given as xx (ej ) =
2 W 25 = j )|2 j + 0.5ej2 |2 |A(e |1 e

(1)

2 where W is the variance of the input sequence.

(a) Determine the dierence equation for generating the AR process when the excitation is white noise. (b) Determine the system function for the whitening lter.

Problem 6.3 An ARMA process has an autocorrelation {xx (m)} whose z-transform is given as xx (z) = 9 (z 1/3)(z 3) , (z 1/2)(z 2) 1 < |z| < 2 2 (2)

(a) Determine the lter H(z) for generating {x(n)} from a white noise input sequence. Is H(z) unique? Explain. (b) Determine a stable linear whitening lter for the sequence {x(n)}.

Problem Set 6

Problem 6.4 Consider the ARMA process generated by the dierence equation x(n) = 1.6x(n 1) 0.63x(n 2) + w(n) + 0.9w(n 1) (a) Determine the system function of the whitening lter and its poles and zeros. (b) Determine the power density spectrum of {x(n)}. (3)

Problem 6.5 An AR(2) process ie dened by the following dierence equation x[n] = x[n 1] 0.6x[n 2] + w[n]
2 where w[n] is a white noise process with variance w . Use the Yule-Walker equation to solve for the values of the autocorrelation xx (0), xx (1), and xx (2).

Problem 6.6 Use the orthogonality principle to determine the normal equations and the resulting minimum MSE for a forward predictor of order N that predicts m samples (m > 1) into the future (m-step forward predictor). Sketch the prediction error lter.

Problem 6.1

Problem 6.2

Problem 6.3

Problem 6.4

Problem 6.5

Problem 6.6

You might also like