Sei sulla pagina 1di 4

University of California, Berkeley

EE 225A: Digital Signal Processing


Problem Set 1

Spring 2015
due Wednesday, January 28, 2015

Reading
Chapter 2 of Statistical Digital Signal Processing and Modeling by Monson Hayes
Problems
1. (Fourier Transform)1
Downsampling a sequence by an integer factor N results in a sequence y[n] =
x[nN ]. Prove that in the Fourier domain
N 1
1 X
Y (e ) =
X(ej(2k)/N ),
N k=0
j

by going back to the underlying time-domain signal and resampling it with an


N -times longer sampling period. That is, consider x[n] and y[n] = x[nN ] as two
sampled versions of the same continuous-time signal, with sampling periods T
and N T , respectively.
Hint: Recall that the discrete-time Fourier transform X(ej ) of x[n] is
j

X(e ) = XT


T



+

1 X
2
=
XC
k
,
T k=
T
T

where T is the sampling period. Then Y (ej ) = XN T ( NT ) (since the sampling


period is now NT), where XN T ( NT ) can be written similarly to the above equation. Finally, split the sum involved in XN T ( NT ) into k = nN + l, and gather
terms.
2. (Impulse Response Inverse)
When Tart Vimpson plugged in his new digital TV, he was dismayed to see ghosts
on the screen. He figured they are created by a secondary delayed RF propagation
path. Some measurements revealed that this problem could be modeled by his
digital video signal being filtered by an LTI system with unit pulse response
h(n) = (n) 0.2(n k0 ).
1

DSP Prelim, Spring 2005

To correct the problem, he wishes to process his signal by an inverse filter with
unit pulse response g(n), so that the effect of h(n) is cancelled.
1
(a) To determine g(n) he computes the N -point DFT {H(k)}N
k=0 , with N = 4k0 ,
of the sequence h(n) and then defines g(n) as the inverse DFT of G(k) =
1/H(k), for k = 0, 1, . . . , N 1. Determine g(n)

(b) Compute the convolution g(n)h(n). Is the system with unit pulse response
g(n) the inverse of the system with unit pulse respone h(n)?
(c) Find the transfer function of an exact inverse gI (n), and comment on the
relation to g(n).
3. (Z-Transform and DFT)
Given a 10-point sequence x[n], we wish to find the equally spaced samples of its
Z-transform X(z) on the contour shown in Figure 1. Show how to achieve this
goal by using the DFT.
Im
z-plane

11
00
00
11
00
11
00
11
00
11
00
11
11
00
00
11
00
11

1
0
0
1
0
1

1
0
0
1
0
1

1
0
0
1
00
11
00
11
00
11
11
00
00
11
00
11
0
1
0
1
0
1

1
0
0
1

2
10
2
20
Re

Circle with
radius = 1/2

Figure 1. Sampled locations of the Z-Transform for Problem 3

4. (Hayes, Problem 2.3)


Find the minimum norm solution to the following set of underdetermined linear
equations,

x
1


 

1 0 2 1
x2 = 1
1 1 0 1 x3
1
x4

5. (Hayes, Problem 2.4)


Consider the set of inconsistent

1
0
1

linear equation Ax = b given by



0 
1
x1

1
= 1
x2
1
0

(a) Find the least squares solution to these equations.


(b) Find the projection matrix PA .
= PA b to b.
(c) Find the best approximation b
(d) Consider the matrix
P
A = I PA

Find the vector b = P


A b and show that it is orthogonal to b. What does

the matrix PA represent?


6. (Eigenvalues and Eigenvectors)


1 2
Consider the matrices A =
2 1
be equal to zero at the same time.


and B =


where and cannot

(a) Give the eigenvalues and eigenvectors for A and B (make sure that the
eigenvectors are of norm 1).
(b) Show that A = V DV T where the columns of V correspond to the eigenvectors of A and D is a diagonal matrix whose main diagonal corresponds to
the eigenvalues of A. Is V a unitary matrix? Why?
(c) Check your results using the built-in Matlab function eig.
(d) Compute the determinants of A and B. Is A invertible? If it is, give its
inverse; if not, say why.
(e) When is the matrix B invertible? Compute B 1 when it exists.
7. (Hayes, Problem 2.19)
Consider the following quadratic function of two real-valued variables z1 and z2 ,
f (z1 , z2 ) = 3z12 + 3z22 + 4z1 z2 + 8.
(a) Show that f is convex by showing its Hessian is positive definite.
(b) Find the values z1 and z2 that minimize f (z1 , z2 ) subject to the constraint
that z1 + z2 = 1 and determine the minimum value of f (z1 , z2 ) under this
constraint.
3

Optional Problems
1. (Periodicity and LTI)

(a) Consider a discrete-time linear time-invariant (LTI) system with impulse


response h[n]. If the input x[n] is a periodic sequence with period N (i.e.
x[n] = x[n + N ]), show that the output y[n] is also a periodic sequence with
period N .
(b) Determine which of the following discrete signals is periodic. If the signal is
periodic, determine its period.
i. x[n] = ej2n/5
ii. x[n] = nejn
iii. x[n] = ejn
2. (Hayes, Problem 2.12)
A necessary and sufficient condition for a Hermitian matrix A to be positive
definite is that there exists a nonsingular matrix W such that
A = W H W.
(a) Prove this result.
(b) Find a factorization of the form W H W for the matrix


2 1
A=
.
1 2
3. (Stability, Causality, and LTI)
For each of the following systems, determine whether the system is (1) stable,
(2) causal, (3) linear, and (4) time invariant.
(a) T [x[n]] = x[n2 ]
P
(b) T [x[n]] = x[n]
k=0 [n k]

Potrebbero piacerti anche