EE123 Digital Signal Processingee123/sp15/Notes/Lecture03_DTFT.pdf · • Read Ch 2 2.0-2.9 (2nd...
Transcript of EE123 Digital Signal Processingee123/sp15/Notes/Lecture03_DTFT.pdf · • Read Ch 2 2.0-2.9 (2nd...
M. Lustig, EECS UC Berkeley
EE123Digital Signal Processing
Lecture 3
M. Lustig, EECS UC Berkeley
A couple of things
• Read Ch 2 2.0-2.9 (2nd edition is fine)• Class webcasted in bcourses.berkeley.edu• Prof. Lustig’s office hours: TBA• Frank Ong
– M 4p-5, 212 Cory (this week I will cover his OH)– Lab Bash – Cancelled this week
• My office hours– Th 5-6pm, 212 Cory
• HW1 due this Friday (1/30), 11:59 pm– Submit on bCourses
• Lab0 due next Friday (2/6)
M. Lustig, EECS UC Berkeley
Discrete Time Systems
• Causality• Memoryless• Linearity• Time Invariance• BIBO stability
M. Lustig, EECS UC Berkeley
Discrete-Time LTI Systems
• The impulse response h[n] completely characterizes an LTI system “DNA of LTI”
discrete convolution
Sum of weighted, delayed impulse responses!
M. Lustig, EECS UC Berkeley
BIBO Stability of LTI Systems
• An LTI system is BIBO stable iff h[n] is absolutely summable
Cool DSP: Steganography
• Hide signals in other signals
• Example: hiding an image in a song
“Secret Message”
0110011…
Encode
0.2267 0.2268 0.2269 0.227 0.2271 0.2272 0.22730
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Time (s)
Am
plitu
de
Secretmessage
?
AudioSignal
Encode
CoverSignal
256x320
Let’s compare the two signals
Overall Signalsx[n] and y[n]look identical
(play the 2 clips)
Restricted signalsx5[n] and y5[n]very different!
0.2267 0.2268 0.2269 0.227 0.2271 0.2272 0.2273-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
Time (s)
Am
plitu
de
Original signal (x[n])Modified Signal (y[n])
= Signal y[n] at 5th
decimal place
0.2267 0.2268 0.2269 0.227 0.2271 0.2272 0.22730
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Time (s)
Am
plitu
de
Original signal (x[n])Modified Signal (y[n])
How should we decode the secret message?
Decode
Rearrange
256x320
Bits toBytes
0 1 2 3 4 5 6 7
x 10-3
0
10
20
30
40
50
60
70
80
90
Time (s)
Byt
e M
essa
ge
Linear? Time-invariant? BIBO?
M. Lustig, EE123 UCBM. Lustig, EE123 UCB
Discrete-Time Fourier Transform (DTFT)Discrete-Time Fourier Transform (DTFT)
Why one is sumand the other integral?
Why use one over the other?Alternative
M. Lustig, EE123 UCBM. Lustig, EE123 UCB
Example 1:Example 1:w[n] “window”
DTFT:
Recall:
N-N
M. Lustig, EE123 UCBM. Lustig, EE123 UCB
Example 1 cont.Example 1 cont.
DTFT:
M. Lustig, EE123 UCBM. Lustig, EE123 UCB
Example 1 cont.Example 1 cont.
DTFT:
j
periodic sinc
-
M. Lustig, EE123 UCBM. Lustig, EE123 UCB
Example 1 cont.Example 1 cont.
from l’Hôpitalalso, Σx[n]
=1, why?
M. Lustig, EE123 UCBM. Lustig, EE123 UCB
Properties of the DTFTProperties of the DTFT
Periodicity:
if x[n] is real
Conjugate Symmetry:
Big deal for: MRI, Communications, more....
M. Lustig, EE123 UCBM. Lustig, EE123 UCB
Half Fourier Imaging in MRHalf Fourier Imaging in MR
k-space (Raw Data) Image
Discrete Fourier transform
Complete based on conjugate symmetryHalf the Scan time!
Complete based on conjugate symmetryHalf the Scan time!
M. Lustig, EE123 UCBM. Lustig, EE123 UCB
SSB ModulationSSB ModulationReal Baseband signal has conjugate symmetric spectrum
AM modulation
SSB-SC reduced power, half bandwidth
M. Lustig, EE123 UCBM. Lustig, EE123 UCB
SSBSSB
http://www.youtube.com/watch?v=y0qi9Fr2j6Y&list=PLA5FE5E811C57CF77
M. Lustig, EE123 UCBM. Lustig, EE123 UCB
Properties of the DTFT cont.Properties of the DTFT cont.
Time-Reversal
If x[n] = x[-n] and x[n] is real, then:
M. Lustig, EE123 UCBM. Lustig, EE123 UCB
Q: Suppose:
A: Decompose x[n] to even and odd functions
M. Lustig, EE123 UCBM. Lustig, EE123 UCB
Oops!
M. Lustig, EE123 UCBM. Lustig, EE123 UCB
Time-Freq Shifting/modulation:
Properties of the DTFT cont.Properties of the DTFT cont.
Good for MRI! Why
M. Lustig, EE123 UCBM. Lustig, EE123 UCB
Example 2Example 2What is the DTFT of:
High Pass Filter
See 2.9 for more properties