Name:

Student No.: University of Toronto Electrical & Computer Engineering ECE 362, Winter 2013 Thursday, April 4, 2013

Test #2
Professor Deepa Kundur Duration: 110 minutes All work must be performed within the space provided. You may use the back of sheets if necessary. No aid sheet allowed. No cell phones, PDAs, etc. allowed. You may use a Type 2 (nonprogrammable) calculator. Answer all questions. If a particular question seems unclear, please explicitly state any reasonable assumptions and proceed with the problem. Please properly label all points of interest on sketches and graphs that you are requested to draw, so that there is no ambiguity. For full marks, show all steps and present results clearly. Important tables are given at the end of the test.

Question 1 2

Earned Grade

/10 /10
3

/10
4

/10
5

/10
Total

/50

1. (10 marks) (a) Assume the z -Transform of x(n) is given by X (z ) with region of convergence (ROC) r1 < |z | < r2 . Prove the time-reversal property of the z-Transform from rst principles: x(n)
Z

X (z 1 ),

1 1 < |z | < . r2 r1

For full points, please show all steps. (b) Consider a signal x(n) with z -Transform: b0 + b1 z 1 + b2 z 2 + b3 z 3 . a0 + a1 z 1 + a2 z 2 + a3 z 3 For what general values of a0 , a1 , a2 , a3 , b1 and b2 is x(n) even? X (z ) =

Solution: (a) We start from the denition of the z -Transform of x(n). Let y (n) = x(n) and m = n.

Y (z ) =
n=

y (n)z

=
n=

x(n)z n x(m)(z 1 )m = X (z 1 )
m=

=
m=

x(m)z m =

For the ROC, where X (z ) was nite, now X (z 1 ) is nite, so we have: r1 < r1 < r1 < 1 < r2 |z | |z 1 |
1 |z |

< r2 < r2 < r2 < 1 r1

|z |

(b) If x(n) is even, then x(n) = x(n). Therefore, X (z ) = X (z 1 ) b0 + b1 z 1 + b2 z 2 + b3 z 3 b0 + b1 z + b2 z 2 + b3 z 3 z 3 b3 + b2 z 1 + b1 z 2 + b0 z 3 = = a0 + a1 z 1 + a2 z 2 + a3 z 3 a0 + a1 z + a2 z 2 + a3 z 3 z 3 a3 + a2 z 1 + a1 z 2 + a0 z 3 b0 + b1 z 1 + b2 z 2 + b3 z 3 b3 + b2 z 1 + b1 z 2 + b0 z 3 = for = 0 a0 + a1 z 1 + a2 z 2 + a3 z 3 a3 + a2 z 1 + a1 z 2 + a0 z 3 Therefore, we may have the following general relationships amongst the parameters: b0 = b3 and b3 = b0 b1 = b2 and b2 = b1 a0 = a3 and a3 = a0 a1 = a2 and a2 = a1

We can see from above that a solution only exists for 2 = 1 or = 1, since for for example the rst set of equalities: b0 = b3 = b0 = 2 b0 . Therefore, the signal is even for: b0 = b3 and b1 = b2 a0 = a3 and a1 = a2 where = 1.

2. (10 marks) Consider the signal: x(n) = { 1, 0, 1, 2, 3 } with discrete-time Fourier transform (DTFT) X ( ) = XR ( ) + jXI ( ) where XR ( ) and XI ( ) are the real and imaginary parts of X ( ) in rectangular coordinates. (a) Determine and sketch the signal y1 (n) with DTFT: Y1 ( ) = XR ( ) + XI ( ). (b) Determine and sketch the signal y2 (n) with DTFT: Y2 ( ) = XR ( ) jXI ( ). (c) Determine and sketch the signal y3 (n) with DTFT: Y3 ( ) = j
2

XR ()XI ( )d.

Solution : From the tables at the end of this test and given x(n) is real, we can see that x(n) + x (n) x(n) + x(n) F = = xe (n) XR ( ) 2 2 x(n) x (n) x(n) x(n) F = = xo (n) jXI ( ) 2 2 jxo (n) XI ( ) We compute the even and odd components of x(n) to give, xe (n) = = xo (n) = = 1 [{1, 0, 1, 2, 3, 0, 0} + {0, 0, 3, 2, 1, 0, 1}] 2 1 { 1 2 , 0, 1, 2, 1, 0, 2 } 1 [{1, 0, 1, 2, 3, 0, 0} {0, 0, 3, 2, 1, 0, 1}] 2 1 { 2 , 0, 2, 0, 2, 0, 1 2 }
F

(a) Therefore, we see that y1 (n) = xe (n) jxo (n) Y1 ( ) = XR ( ) + XI ( ) to give y1 (n) = {
1 2 1 j2 , 0, 1 + j 2, 2, 1 j 2, 0, 1 2 F

+ j1 2

(b) Similarly, we see that y2 (n) = xe (n) xo (n) Y2 ( ) = XR ( ) jXI ( ) to give y2 (n) = { 0, 0, 3, 2, 1, 0, 1 } (c) Again, we see that y3 (n) = 2 xe (n) xo (n) Y3 ( ) = j
2 F F

XR ()XI ( )d

to give y3 (n) = {
2,

0, 4, 0, 4, 0, 2

3. (10 marks) Consider the following system for L Z+ : y (n) = x(n + L) x(n L) (1)

(a) Using the following standard building block elements: adders, constant multipliers, unit delay elements, unit advance elements and signal multipliers, determine a realization of the system that uses minimum unit delay/advance elements. (b) Determine and sketch the magnitude and phase of the frequency response of the system Eq. 1. (c) Determine the locations of all spectral nulls, if any. (d) This system is not causal. Suppose you connect the following system y (n) = x(n M ) with input x(n), output y (n) and M Z in series with the system of Eq. 1. What values of M would make the new overall series connection system causal? For the minimum M , determine the input-output equation of this overall series connected causal system. (e) Determine the frequency response expression of the system in part (d).

Solution : (a) The realization is given below where there are L unit delay and unit advance elements in each row.

(b) The frequency response is given by: y (n) = x(n + L) x(n L) Y ( ) = ejL X ( ) ejL X ( ) = [ejL ejL ]X ( ) Y ( ) H ( ) = = [ejL ejL ] = 2j sin(L) X ( ) |H ( )| = 2| sin(L)| H ( ) =
2

(2k+1) 2k L < < L (2k1) 2k < < L L

Adder:

... ...

+ -

,k Z

(c) The spectral nulls occur when H ( ) = 2j sin(L) = 0, which occurs when L = k or = k L for k Z. (d) The overall series connection would be causal for M L, the minimum occurring for M = L. The overall input-output equation is given by:

y1 (n) = x1 (n + L) x1 (n L) y2 (n) = x2 (n M ) = y1 (n M ) = x1 (n M + L) x1 (n M L) y (n) = x(n) x(n 2L) (e) We have that H1 ( ) = 2j sin(L) H2 ( ) = ejL Hoverall ( ) = H1 ( ) H2 ( ) = 2jejL sin(L) for M = L

4. (10 marks) (a) Compute the N -point discrete Fourier transform (DFT) of the signal: x1 (n) = (n). (b) Show

(n + lN ) =
l=

1 N

N 1

ej (2/N )kn .
k=0

Solution: (a) The N -point discrete Fourier transform (DFT) of x1 (n) = (n) is:
N 1

X (k ) =
k=0 N 1

x(n)ej (2/N )kn , k = 0, 1, . . . , N 1 (n)ej (2/N )kn

k=0

= = 1,

k = 0, 1, . . . , N 1

(b) We see that l= (n + lN ) is a periodic repetition of (n) for period N . Therefore, its discrete-time Fourier series (DTFS) coecients are ck = 1 for all k Z (since the DFT is just a windowed version of the DTFS such that the DTFS is a periodic repetition of the DFT for k = 0, 1, . . . , N 1). The DTFS synthesis expression is given by: x(n) = 1 N
N 1

ck ej (2/N )kn
k=0

We also know that:

(n + lN )
l=

DTFS

which we can plug back in to the DTFS synthesis equation to give:

(n + lN ) =
l=

1 N

N 1

ej (2/N )kn .
k=0

5. (10 marks) Since the class seemed concerned about drawing butteries, the buttery representation for a radix-2 decimation-in-time 8-point FFT is provided below. Please ll in the blanks (i.e., the slightly shaded boxes, specically representing the input values into the buttery network and the four twiddle factors between Stage 2 and Stage 3). Hint: the missing twiddle factors can be deduced by taking a rst-level radix-2 decimation-in-time decomposition that we did in the lectures and nding the WN factor that comes out of one of the summations. Here is how you would start:
N 1

X (k ) =
n=0

kn x(n)WN = n even

kn x(n)WN + n odd

kn x(n)WN

k = 0, 1, . . . , N 1

Solution:

N 1

X (k ) =
n=0

kn x(n)WN

k = 0, 1, . . . , N 1
kn x(n)WN n odd (N/2)1

=
n even

kn x(n)WN + (N/2)1

=
m=0 (N/2)1

x(2m)WN

k(2m)

+
m=0 (N/2)1

x(2m + 1)WN

k(2m+1)

=
m=0 (N/2)1

2km x(2m) WN + f1 (m) m=0

2km k x(2m + 1) WN WN f2 (m) km f2 (m)WN/ 2 m=0

N 2

(N/2)1 km k f1 (m)WN/ 2 +WN

=
m=0
N 2

DFT of f1 (m) k WN F2 (k ),

DFT of f2 (m)

= F1 (k ) +

k = 0, 1, . . . , N 1

x(0) x(0) x(4) x(4) x(2) x(2) x(6) x(6) x(1) x(1) x(5) x(5) x(3) x(3) x(7) x(7)

Stage 1

Stage 2

Stage 3

X(0) X(1)

W8

-1 W8 -1 W8
2 0

X(2) X(3)
0 W W8 8 0

W8

-1

-1

-1
1 W W8 8 1

X(4) X(5) X(6) X(7)

W8

-1 W8 -1 W8
2 3 W W8 8 3 0 2 W W8 8 2

-1

-1

W8

-1

-1

-1

10

Common z-Transform Pairs

1 2 3 4 5 6 7 8 9 10 Signal, x(n) (n) u(n) an u(n) nan u(n) an u(n 1) nan u(n 1) cos(0 n)u(n) sin(0 n)u(n) a cos(0 n)u(n) an sin(0 n)u(n)
n

z -Transform, X (z ) 1
1 1z 1 1 1az 1 az 1 (1az 1 )2 1 1az 1 az 1 (1az 1 )2 1z 1 cos 0 12z 1 cos 0 +z 2 z 1 sin 0 12z 1 cos 0 +z 2 1az 1 cos 0 12az 1 cos 0 +a2 z 2 1az 1 sin 0 12az 1 cos 0 +a2 z 2

ROC All z |z | > 1 |z | > |a| |z | > |a| |z | < |a| |z | < |a| |z | > 1 |z | > 1 |z | > |a| |z | > |a|

z-Transform Properties
Property Notation: Time Domain x(n) x1 (n) x2 (n) a1 x1 (n) + a2 x2 (n) x(n k) z -Domain X (z ) X1 (z ) X2 (z ) a1 X1 (z ) + a2 X2 (z ) z k X (z ) ROC ROC: r2 < |z | < r1 ROC1 ROC2 At least ROC1 ROC2 At least ROC, except z = 0 (if k > 0) and z = (if k < 0) |a|r2 < |z | < |a|r1 1 < |z | < r1 r1 2 ROC r 2 < |z | < r 1 At least ROC1 ROC2

Linearity: Time shifting:

z -Scaling: Time reversal Conjugation: z -Dierentiation: Convolution:

an x(n) x(n) x (n) n x(n) x1 (n) x2 (n)

X (a1 z ) X (z 1 ) X (z ) (z ) z dX dz X1 (z )X2 (z )

DTFT Theorems and Properties

Property Notation: Time Domain x(n) x1 (n) x2 (n) a1 x1 (n) + a2 x2 (n) x (n k ) x(n) x1 (n) x2 (n) x1 (n)x2 (n) rx1 x2 (l) = x1 (l) x2 (l) rxx (l) = x(l) x(l) Frequency Domain X ( ) X1 ( ) X1 ( ) a1 X1 ( ) + a2 X2 ( ) ejk X ( ) X ( ) X1 ( )X2 ( ) 1 X1 ()X2 ( )d 2 2 Sx1 x2 ( ) = X1 ( )X2 ( ) = X1 ( )X2 ( ) [if x2 (n) real] Sxx ( ) = |X ( )|2

Linearity: Time shifting: Time reversal Convolution: Multiplication: Correlation: Wiener-Khintchine:

11

DTFT Symmetry Properties

Time Sequence x(n) x (n) x (n) x(n) xR (n) jxI (n) DTFT X ( ) X ( ) X ( ) X ( ) 1 Xe ( ) = 2 [X ( ) + X ( )] 1 Xo ( ) = 2 [X ( ) X ( )] X ( ) = X ( ) XR ( ) = XR ( ) XI ( ) = XI ( ) |X ( )| = |X ( )| X ( ) = X ( ) XR ( ) jXI ( )

x(n) real

1 xe (n) = 2 [x(n) + x (n)] 1 xo (n) = 2 [x(n) x (n)]

DFT Properties
Property Notation: Periodicity: Linearity: Time reversal Circular time shift: Circular frequency shift: Complex conjugate: Circular convolution: Multiplication: Parsevals theorem: Time Domain x(n) x(n) = x(n + N ) a1 x1 (n) + a2 x2 (n) x(N n) x((n l))N x(n)ej 2ln/N x (n) x1 (n) x2 (n) x1 (n)x2 (n) N 1 n=0 x(n)y (n) Frequency Domain X (k ) X (k ) = X (k + N ) a1 X1 (k) + a2 X2 (k) X (N k ) X (k)ej 2kl/N X ((k l))N X (N k ) X1 (k)X2 (k) 1 X1 (k) X2 (k) N N 1 1 k=0 X (k )Y (k ) N

12

Note: The following tables are courtesy of Ashish Khisti and Ravi Adve and were developed originally for ECE355. Fourier Properties
Property Synthesis Analysis
1 N

DTFS x[n] = jk0 n k=<N > ak e ak = jk0 n n=<N > x[n]e x[n] + y [n] ak + bk
1 T

CTFS
k=

DTFT
1 2

CTFT
1 2

x(t) = ak ejk0 t

x[n] = X (ej )ej n d 2

x(t) = X (j )ejt d

ak = x(t)ejk0 t dt

X (ej ) = x[n]ej n
j j

X (j ) = x(t)ejt dt

Linearity Time Shifting Frequency Shift Conjugation Time Reversal Convolution

x(t) + y (t) ak + bk x(t t0 ) ak e

jk0 t0

x[n] + y [n] X (e ) + Y (e
j n0

x(t) + y (t) )
j

X (j ) + Y (j ) ) x(t t0 ) ejt0 X (j ) x(t)ej0 t X (j ( 0 )) x (t) X (j ) x(t) X (j )

j

x[n n0 ] ak e

j 2n0 k/N

x[n n0 ] e

X (e

x[n]ej 2mn/N akm x [n] a k x[n] ak

N 1 r =0

x(t)ejm0 t akm x (t) a k x(t) ak

T

x[n]ej 0 n X (ej (0 )n ) x [n] X (ej ) x[n] X (ej ) x[n] y [n] X (e

j

x[r]y [n r] N a k bk
N 1 r =0

x( )y (t )d T ak bk

)Y (e

x(t) y (t) X (j )Y (j ) x(t)y (t)

1 X (j ) 2 dx(t) dt

x[n]y [n] Multiplication

ar bkr

x(t)y (t) ak bk
1 2 2

x[n]y [n] X (ej )Y (ej () )d

Y (j )

First Dierence/ Derivative Running Sum/ Integration Parsevals Relation Real and even signals Real and odd signals

x[n] x[n 1] (1 ej 2k/N )ak

n k= x[k ] ak 1ej 2k/N 1 N N 1 2 n=0 |x[n]| N 1 2 k=0 |ak |

dx(t) dt

jk0 ak
ak jk0

x[n] x[n 1] (1 ej )X (ej )

n k= X (e ) x[k] 1 ej j0 +X (e ) () n=
j

jX (j )

x( )d

(j ) x( )d Xj +X (j 0) ( ) |x(t)|2 dt |X (j )|2 d

1 T

T k=

|x(t)|2 dt |ak |2

1 2

|x[n]|2 |X (ej )|2 d 2

1 2

Real and even in frequency domain Purely imaginary and odd in frequency domain

Additional Property: A real-valued time-domain signal x(t) or x[n] will have a conjugate-symmetric Fourier representation. Notes: 1. For the CTFS, the signal x(t) has a period of T , fundamental frequency 0 = 2/T ; for the DTFS, the signal x[n] has a period of N , fundamental frequency 0 = 2/N . ak and bk denote the Fourier coecients of x(t) (or x[n]) and y (t) (or y [n]) respectively. 2. Periodic convolutions can be evaluated by summing or integrating over any single period, not just those indicated above. 3. The Running Sum formula for the DTFT above is valid for in the range < .

13

Fourier Pairs
Fourier Series Coefficients of Periodic Signals Continuous-Time Time Domain x(t) Aej0 t A cos(0 t) A sin(0 t) x(t) = A
n=

Discrete-Time Time Domain x[n] Aej 0 n A cos(0 n) A sin(0 n) x[n] = A

k=

Frequency Domain ak a1 = A ak = 0, k = 1 a1 = a1 = A/2 ak = 0, k = 1

A a1 = a 1 = 2j ak = 0, k = 1

Frequency Domain ak a1 = A, ak = 0, k = 1 a1 = a1 = A/2 ak = 0, k = 1

A a1 = a 1 = 2j ak = 0, k = 1

a0 = A, ak = 0 otherwise ak = a0 =
1 T

a0 = A, ak = 0 otherwise ak =
1 N

(t nT )

[n kN ]

Periodic square wave x(t) = 1 0 |t | < T 1 T1 < |t|

T 2

2T1 T sin(k0 T1 ) ,k=0 ak = k

and x(t) = x(t + T ) Fourier Transform Pairs Continuous-Time Time Domain x(t) x(t) = 1 , |t | < T 1 0 , |t | > T 1 sin W t t (t ) 1 u(t) eat u(t), Re(a) > 0 tn1 at e u(t), Re(a) > 0 (n 1)!

Discrete-Time Time Domain x[n] x[n] = 1, |n| N1 0, |n| > N1 sin W n n [n] 1 u[n] an u[n], |a| < 1 (n + r 1)! n a u[n], |a| < 1 n!(r 1)! Frequency Domain X (ej ) sin((N1 + 1/2)) sin(/2) 1, || W j X (e ) = 0, otherwise 1 2 () 1 + () 1 ej 1 1 aej 1 (1 aej )r

Frequency Domain X (j ) 2 sin(T1 ) 1, | | < W X (j ) = 0, otherwise 1 2 ( ) 1 + ( ) j 1 a + j 1 (a + j )n

In the Fourier series table, 0 = 2 and 0 = 2 , where T and N are the periods of x(t) and x[n] respectively. T N For the DTFS, ak is given only for k in the range N/2 + 1 k N/2 for even N , (N 1)/2 k (N 1)/2 for odd N , and ak = ak+N ; for the DTFT X (ej ) is given only for in the range < , and X (ej ) = X (ej (+2) ).

Fourier Transform for Periodic Signals:

x(t) =
k=

ak ejk0 t X (j ) = 2
k=

ak ( k0 )

x[n] =
k=<N >

ak ejk0 n X (ej ) = 2
k=

ak ( k0 )

14