DSP MATLAB Assignments
Uploaded by
Akshansh ChaudharyDSP MATLAB Assignments
Uploaded by
Akshansh ChaudharyDigital Signal Processing MATLAB Notes, First Edition
Copyright 2013 Akshansh
ALL RIGHTS RESERVED.
Presented by:
Akshansh Chaudhary
Graduate of BITS Pilani, Dubai Campus
Batch of 2011
Course content by:
Dr. R. Mary Lourde
Then Faculty, BITS Pilani, Dubai Campus
Layout design by:
AC Creations 2013
The course content was prepared during Fall, 2013.
More content available at: www.Akshansh.weebly.com
DISCLAIMER: While the document has attempted to make the information as accurate as possible, the information on this document is for personal and/or
educational use only and is provided in good faith without any express or implied warranty. There is no guarantee given as to the accuracy or currency of any
individual items. The document does not accept responsibility for any loss or damage occasioned by use of the information contained and acknowledges
credit of author(s) where ever due. While the document makes every effort to ensure the availability and integrity of its resources, it cannot guarantee that
these will always be available, and/or free of any defects, including viruses. Users should take this into account when accessing the resources. All access and
use is at the risk of the user and owner reserves that right to control or deny access.
Information, notes, models, graph etc. provided about subjects, topics, units, courses and any other similar arrangements for course/paper, are an expression
to facilitate ease of learning and dissemination of views/personal understanding and as such they are not to be taken as a firm offer or undertaking. The
document reserves the right to discontinue or vary such subjects, topic, units, courses, or arrangements at any time without notice and to impose limitations
on accessibility in any course.
MATLAB BASICS
Digital Signal Processing
Akshansh Chaudhary
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
MATLAB Assignment 1
Digital Signal Processing
Akshansh Chaudhary
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
2011AAPS300U
ks
ha
ns
Digital Signal Processing Lab
Assignment 1
Butterworth filter
AT
LA
(D
ig
i
ta
lS
ig
na
lP
ro
ce
ss
in
g)
A
Instructions [N1,Wn1]=buttord([1200 1800],[800 2200],0.5,30,'s')
[B1,A1]=butter(N1,Wn1,'s')
[h1,f1]=freqs(B1,A1)
plot(f1,h1)
plot(f1,20*log10(abs(h1)))
Akshansh Chaudhary
Akshansh Chaudhary
LA
AT
M
(D
g)
A
in
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
AT
M
(D
g)
A
in
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Chebyshev's filter
Instructions -
ks
ha
ns
[N2,Wn2]=cheb1ord([1200 1800],[800 2200],0.5,30,'s')
[B2,A2]=cheby1(4,0.5,[1200 1800],'s')
[h2,f2]=freqs(B2,A2)
plot(f2,h2)
AT
LA
(D
ig
i
ta
lS
ig
na
lP
ro
ce
ss
in
g)
A
plot(f2,20*log10(abs(h2)))
Akshansh Chaudhary
Akshansh Chaudhary
LA
AT
M
(D
g)
A
in
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
AT
M
(D
g)
A
in
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Elliptical filter
Instructions -
ks
ha
ns
[N3,Wn3]=ellipord([1200 1800],[800 2200],0.5,30,'s')
[B3,A3]=ellip(N3,0.5,30,[1200 1800],'s')
[h3,f3)=freqs(B3,A3)
plot(f3,h3)
AT
LA
(D
ig
i
ta
lS
ig
na
lP
ro
ce
ss
in
g)
A
plot(f3,20*log10(abs(h3)))
Akshansh Chaudhary
Akshansh Chaudhary
LA
AT
M
(D
g)
A
in
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
AT
M
(D
g)
A
in
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Comparison
AT
LA
(D
ig
i
ta
lS
ig
na
lP
ro
ce
ss
in
g)
A
ks
ha
ns
As the order decreases, frequency decreases.
For Butterworth filter, n=7.
For Chebyshevs filter, n=4
For Elliptical Filter, n=3.
Akshansh Chaudhary
MATLAB Assignment 2
Digital Signal Processing
Akshansh Chaudhary
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
ro
ce
ss
in
g)
A
(a) H(z)=(z2-z)/(z2-0.9051z+0.4096)
Determine the location of poles and zeroes.
Plot the pole zero map of function.
(b) Repeat the same if function is given by
H(z)=(1-z-1)/1-0.9051z-1+0.4096z-2)
(c) Plot pole-zero diagram of H(z) in both cases
(i.)
With numerator and denominator polynomial coefficients as inputs
(ii.)
With poles and zeroes as inputs
(d) Plot the frequency response of the DT system with a sampling frequency of
(i.)
1 kHz
(ii.)
10kHz
(iii.)
100kHz
ks
ha
ns
Question: The transfer function of a discrete time system is given by
Digital Signal Processing
MATLAB Assignment 2
Akshansh Chaudhary
Id no. 2011AAPS300U
lP
Solution:
na
Let the transfer functions be named as
ig
(a) TF1 = (z2-z)/(z2-0.9051z+0.4096)
(b) TF2 = (1-z-1)/1-0.9051z-1+0.4096z-2)
lS
Subpart 1: Polynomial coefficients of TF1 and TF2 as inputs:
M
AT
LA
(D
ig
ita
a. For TF1,
Commands:
A=[1 -1 0]; B=[1 -0.9051 0.4096];
pzmap(A,B)
From this, we get the following graph
Akshansh Chaudhary
lP
Subpart 2: Poles and Zeroes as inputs:
ro
ce
ss
in
g)
A
ks
ha
ns
b. For TF2,
Commands:
num=[1 -1]; den=[1 -0.9051 0.4096];
pzmap(num,den)
We get the following graph
M
AT
LA
(D
ig
ita
lS
ig
na
a. For TF1,
Finding the poles and zeroes of the TF.
Commands:
>> A1=roots(A)
A1 =
0
1
>> B1=roots(B)
B1 =
0.4526 + 0.4525i
0.4526 - 0.4525i
b.
For TF2,
Commands:
>> num1=roots(num)
num1 =
1
>> den1=roots(den)
den1 =
0.4526 + 0.4525i
0.4526 - 0.4525i
Akshansh Chaudhary
ro
ce
ss
in
g)
A
ks
ha
ns
Now, commands for making the pole zero plot for Subpart 2:
Commands:
a. For TF1
>> pzmap(B1,A1)
lP
b. For TF2
M
AT
LA
(D
ig
ita
lS
ig
na
>> pzmap(den1,num1)
Akshansh Chaudhary
Part (d). Finding frequency response of DT system with a sampling frequency of
1kHz
Command: freqz(num,den,1000)
ii.
10kHz
Command: freqz(num,den,10000)
M
AT
LA
(D
ig
ita
lS
ig
na
lP
ro
ce
ss
in
g)
A
ks
ha
ns
i.
Akshansh Chaudhary
100kHz
Command:
Freqz(num,den,100000)
na
lP
ro
ce
ss
in
g)
A
ks
ha
ns
iii.
ig
Observation and Analysis
lS
It was found that the order of the given transfer function (in z domain) changes when the transfer
function is rearranged from positive powers of z to negative powers of z.
ita
So, TF1 (positive powers of z) has order =2 (w.r.t numerator)
(D
ig
And, TF2 (negative powers of z) has order =1 (w.r.t numerator)
Also, it was found that the pole zero plot using the coefficients of the transfer function as matrix and
the poles and zeroes as matrix is same for both TF1 and TF2.
LA
Finally,
M
AT
In the frequency response plots, the plots for both TF1 and TF2 were found to be the same. And, on
seeing the plots on different frequencies, the plots were nearly the same with little difference in the
magnitude part of the plot on changing the frequency.
Akshansh Chaudhary
MATLAB Assignment 3
Digital Signal Processing
Akshansh Chaudhary
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
Akshansh Chaudhary
LA
B
AT
M
(D
in
g)
A
ss
ce
ro
lP
na
ig
lS
ta
ig
i
ks
ha
ns
sh
a
ns
h
Assignment 3
Akshansh Chaudhary
ID 2011AAPS300U
Dated - 7.10.2013
Question-
ng
)-
Ak
Given the specifications of a band stop filter, make it an analog filter and convert it to
digital filter using impulse invariant method and Bilinear z transform method. Also, find
the magnitude plot, phase plot, group delay and pole zero diagrams for both the methods.
si
Solution -
ig
na
lP
ro
c
es
Part (1)
Making analog elliptical
filter
ig
ita
lS
Initializing the value of Wp,Ws,Rp, Rs for the elliptical
filter
AL
TA
(D
Rp=0.2
Rp =
0.2000
>> Rs=40
Rs =
40
>> Fs=100000
Fs =
100000
>> Wp=[15000/50000, 30000/50000]
Wp =
0.3000 0.6000
ns
h
>> Ws=[20000/50000,25000/50000]
Ws =
0.4000 0.5000
sh
a
Instructions after initialization
AL
TA
(D
ig
ita
lS
ig
)ng
si
es
ro
c
lP
na
[B1,A1]=ellip(N1,Rp,Rs,Wn1,'stop')
B1 =
Columns 1 through 3
0.3198 -0.4313 1.3982
Columns 4 through 6
-1.2748 2.1536 -1.2748
Columns 7 through 9
1.3982 -0.4313 0.3198
A1 =
Columns 1 through 3
1.0000 -0.9838 1.9828
Columns 4 through 6
-1.4369 1.8869 -0.8671
Columns 7 through 9
0.7043 -0.2039 0.1459
Ak
>> [N1,Wn1]=ellipord(Wp,Ws,Rp,Rs,'s')
N1 =
4
Wn1 =
0.3333 0.6000
sh
a
Ak
ng
)-
(i.) Impulse Invariant Method
ns
h
Part (2)
Converting to Digital
lS
ig
na
lP
ro
c
es
si
[bz,az]=impinvar(B1,A1,Fs)
bz =
1.0e-003 *
Columns 1 through 8
0.0029 -0.0238 0.0859 -0.1772 0.2284 -0.1882 0.0968 -0.0284
Column 9
0.0037
az =
Columns 1 through 8
1.0000 -8.0000 28.0000 -56.0001 70.0002 -56.0002 28.0001 -8.0000
Column 9
1.0000
(D
ig
ita
(a.) Magnitude Plot
TA
>> [H,F]=freqs(bz,az)
>> plot(F,20*log10(abs(H)))
>> xlabel('freq. (in kHz)')
>> ylabel('magnitude (in dB)')
AL
(Note: Graphs got in one version of MATLAB may differ in other versions)
ns
h
sh
a
Ak
)ng
si
es
ro
c
lP
na
AL
TA
(D
ig
ita
lS
ig
Observations from the graph:
The stop band is between 0.2 and 0.25 kHz as required.
Pass band is between 0-0.15 kHz and then 0.3-0.5 kHz as in problem.
The pass band ripple is nearly 0.2 (after zooming the pass band area)
The stop band ripple has a little deviation. Its coming as 30 db (instead of 40 dB given.)
The problem becomes more ideal when the sampling frequency is increased.
(b.) Phase delay
ita
lS
ig
na
lP
ro
c
es
si
ng
)-
Ak
sh
a
ns
h
>> phasedelay(bz,az,512)
ig
(c.) Group Delay
AL
TA
(D
>> grpdelay(bz,az,512)
>>
TA
AL
)-
ng
si
es
ro
c
lP
na
ig
lS
ita
ig
(D
Ak
sh
a
ns
h
TA
AL
)-
ng
si
es
ro
c
lP
na
ig
lS
ita
ig
(D
Ak
ns
h
zplane(bz,az)
sh
a
(d.) Pole zero diagrams
ns
h
sh
a
Ak
)ng
si
es
ro
c
lP
na
ig
lS
ig
ita
(ii.) BZT Method
AL
TA
(D
>> [b1,a1]=bilinear(B1,A1,Fs)
b1 =
Columns 1 through 8
0.3652 -2.9218 10.2263 -20.4526 25.5658 -20.4527 10.2263 -2.9218
Column 9
0.3652
a1 =
Columns 1 through 8
1.0000 -8.0000 28.0000 -56.0001 70.0002 -56.0002 28.0001 -8.0000
Column 9
(a) Magnitude Plot
(D
ig
ita
lS
ig
na
lP
ro
c
es
si
ng
)-
Ak
sh
a
ns
h
>> [H,F]=freqz(B1,A1,512,Fs)
>> plot(F,20*log10(abs(H)))
>> ylabel('magnitude response in dB')
>> xlabel('freq in Hz')
AL
TA
(b.) Phase Plot
>> phasedelay(b1,a1,512)
>>
TA
AL
)-
ng
si
es
ro
c
lP
na
ig
lS
ita
ig
(D
Ak
sh
a
ns
h
TA
AL
)-
ng
si
es
ro
c
lP
na
ig
lS
ita
ig
(D
Ak
sh
a
ns
h
(c.) Group Delay
>> grpdelay(b1,a1,512)
TA
AL
)-
ng
si
es
ro
c
lP
na
ig
>> zplane(b1,a1)
lS
ita
ig
(D
(d.) Pole Zero Plot
Ak
sh
a
ns
h
TA
AL
)-
ng
si
es
ro
c
lP
na
ig
lS
ita
ig
(D
Ak
sh
a
ns
h
sh
a
(Seeing in s domain)
Ak
>> [N1,Wn1]=ellipord(Wp,Ws,Rp,Rs,'s')
>> [B1,A1]=ellip(N1,Rp,Rs,Wn1,'stop')
ng
)-
>> [H,F]=freqs(B1,A1)
AL
TA
(D
ig
ita
lS
ig
na
lP
ro
c
es
si
>>plot(F,abs(H))
>> plot(F, 20*log10(abs(H)))
>> ylabel('H(jw) (magnitude) in dB')
>> xlabel('frequency (Hz)')
ns
h
Showing that we have Band Stop Filter
ns
h
sh
a
Ak
)ng
si
es
ro
c
ita
lS
ig
na
lP
The figure shows the magnitude plot for elliptical filter with the given specifications in
the ANALOG DOMAIN.
From the figure, it is visible that its a band stop filter.
(D
ig
Conclusion
AL
TA
In the analog domain, the sampling interval is from 0 to infinity.
But, as we change it to the digital domain, the sampling interval changes form 0 to the
sampling frequency. Basically, this reduces the sampling interval, or, in a way,
compresses it.
In such a condition, we find distortion in the magnitude, phase, group and phase delay
plots.
This effect is called as Nyquist effect.
Matlab Assignment 4
Digital Signal Processing
Akshansh Chaudhary
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
ig
na
lP
ro
c
es
si
ng
)-
Ak
sh
a
ns
h
Assignment 4
(D
ig
ita
lS
In the given problem, it has been told that the pass band frequency is given as 200 Hz and
stop band frequency is given as 500 Hz.
Now, as its told to design a band pass filter, are left to assume it as:
a. A low pass filter with pass band edge frequency of 200 Hz and stop band edge
frequency of 500 Hz
b. A band pass filter with lower stop band frequency of 0 Hz, lower pass band
frequency of 200 Hz and upper stop band frequency of 500 Hz.
Note: it is left to the designer to choose how to design the filter using the specs given.
AL
TA
Now, assuming the case, b. we assume:
Wsl= 0 Hz
Wpl= 200 Hz
That means, the transition width (Wpl-Wsl)= Delta f (denormalised)=200 Hz
Wsu=500 Hz
So, Wpu= Wsu-200 Hz = 300 Hz
AL
TA
(D
ig
ita
lS
ig
na
lP
ro
c
es
si
ng
)-
Ak
sh
a
>> fs=2000;
>> fn=fs/2; % fn= Nyquist freq.
>> N=7;
>> fc1=100/fn; % 100, as its Wpl - (delta f/2); and /fn, as normalising w.r.t nyquist
>> fc2=400/fn; % 400, as its Wpu + (delta f/2)
>> FC=[fc1 fc2];
>> hn=fir1(N-1, FC, hamming(N));
>> [H,f]=freqz(hn, 1,512,fs);
>> mag=20*log10(abs(H));
>> plot(f,mag,'red');
grid on;
xlabel('freq(Hz)')
>> ylabel('Magnitude response (db)')
>>
ns
h
Commands:
lP
ro
c
es
si
ng
)-
Ak
sh
a
ns
h
Applying the same commands for N=8, we get the following graph
AL
TA
(D
ig
ita
lS
ig
na
Applying the same commands for N=22, we get the following graph
ig
na
lP
ro
c
es
si
ng
)-
Ak
sh
a
ns
h
Proceeding similarly and then, holding all the graphs together, we see:
Blue: N=7
Green: N=8
Red: N=22
Magenta: N=33
AL
TA
(D
ig
ita
lS
Note: finding N by using delta f formula for hamming window
Delta f = 3.3/N
Where Delta f = 200/fs
Using this, we get N=33
ns
h
sh
a
Ak
)ng
si
es
ro
c
ig
na
lP
Clearly, here fpl=2kHz and fpu=4kHz.
For the given problem, assume the sampling frequency as 10kHz, as its not mentioned.
(We know that for denormalizing using Nyquist theorem, we have to divide the
frequencies by (fs/2), where fs is the sampling frequency.
Also, the sampling frequency should be greater than 4kHz.)
lS
For hamming window
AL
TA
(D
ig
ita
N=41;
>> fp1=0.3196; % value got after normalizing it w.r.t Nyquist, and using smearing effect
>> fp2=0.8804;
>> fs=10000;
>> fp=[fp1 fp2];
>> hn=fir1(N-1,fp,hamming(N));
>> hn=fir1(N-1,fp,hamming(N));
>> [H,f]=freqz(hn,1,512,fs);
>> mag=20*log10(abs(H));
>> plot(f,mag,'magenta'); grid on;
>>
ns
h
sh
a
Ak
)ng
si
es
ro
c
na
lP
Clearly, it is visible that the pass band is between 2000 to 5000 Hz (nearly)
lS
ig
For rectangular window
AL
TA
(D
ig
ita
>> fc1=0.37804;
>> fc2=0.81756;
>> fc=[fc1 fc2];
>> hn2=fir1(N-1,fc,rectwin(N));
>> [H2,f2]=freqz(hn2,1,512,fs);
>> mag2=20*log10(abs(H2));
>> plot(f2,mag2,'green');grid on;
ns
h
sh
a
Ak
)ng
es
si
lP
ro
c
It is also having pass band in 2000 to 5000 Hz.
AL
TA
(D
ig
ita
lS
ig
na
Including both graphs
AL
TA
(D
ig
ita
lS
ig
na
lP
ro
c
es
si
ng
)-
Ak
sh
a
ns
h
Inference:
It can be clearly seen that increasing the no. of samples (N)
makes the frequency response more ideal.
Moreover, the graph compresses as N is increased.
Matlab Assignment 5
Digital Signal Processing
Akshansh Chaudhary
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
MATLAB Assignment 6
Digital Signal Processing
Akshansh Chaudhary
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Assignment 6
Q. 7.32
AL
TA
(D
ig
ita
lS
ig
na
lP
ro
c
es
si
ng
)-
Ak
sh
a
ns
h
1.
fpl=13/48;
>> fpu=27/48;
>> fsl=12/48;
>> fsu=15/48;
>> fc=[fpl fpu];
>> N1=53;
>> hd=fir1(N1-1, fc, boxcar(N1));
>> wn=hamming(N1);
>> hn1=fir1(N1-1, fc, wn);
>> fs=48000;
>> [H1,f1]=freqz(hn1, 1, 512, fs);
>> mag1=20*log10(abs(H1));
>> plot(f1, mag1), grid on;
ita
lS
ig
na
lP
ro
c
es
si
ng
)-
Ak
sh
a
ns
h
>> impz(hn1,1)
>>
AL
TA
(D
ig
2.
N2=58;
>> beta=5.65;
>> hn2=fir1(N2-1, fc, kaiser(N2, beta));
>> [H2, f2]=freqz(hn2, 1, 512, fs);
>> mag2=20*log10(abs(H2));
>> plot(f2, mag2, red), grid minor;
TA
AL
)-
ng
si
es
ro
c
lP
na
ig
lS
ita
ig
(D
Ak
sh
a
ns
h
TA
AL
)-
ng
si
es
ro
c
lP
na
ig
lS
ita
ig
(D
Ak
sh
a
ns
h
ns
h
sh
a
Ak
)ng
si
es
ro
c
lP
na
ig
AL
TA
(D
ig
ita
lS
3.
> F=[5000, 8000, 12000, 15000];
>> M=[0 1 0];
>> dp=0.01;
>> ds=0.001;
>> dev=[ds dp ds];
>> [N3, F0, M0, W]=remezord(F, M, dev, fs);
>> [b delta]=remez(N3, F0, M0, W);
>> [H3, f3]=freqz(b, 1, 1024, fs);
>> mag3=20*log10(abs(H3));
>> plot(f3, mag3, green), grid minor;
AL
TA
(D
ig
ita
lS
ig
na
lP
ro
c
es
si
ng
)-
Ak
sh
a
ns
h
Plotting all freq. resaponse on same graph
MATLAB Assignment 7
Digital Signal Processing
Akshansh Chaudhary
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Akshansh Chaudhary
TA
AL
(D
g)
in
ss
ce
ro
lP
na
ig
ita
lS
ig
ks
-A
ha
ns
Assignment 7
Decimation and Interpolation
ns
h
9.12.13
TA
(D
ig
ita
lS
ig
na
Ak
)ng
si
es
ro
c
lP
>> Fs=5000;
>> A=2;
>> B=1;
>> f1=50;
>> f2=100;
>> t=0:1/Fs:1;
>> x=A*cos(2*pi*f1*t)+B*cos(2*pi*f2*t);
>> subplot(3,1,1)
>> stem(x(1:1000))
>> xlabel('Discrete time, nT')
>> ylabel('Input signal level')
>>
>>
>> %Now, decimating for part b.
>> y=decimate(x,10);
>> subplot(3,1,2)
>> stem(y(1:100))
>> xlabel('Discrete time, nT/10')
>> ylabel('Decimated output signal level')
>>
>> %Now, interpolating for part c.
>> y1=interp(y,4);
>> subplot(3,1,3)
>> stem(y1(1:400))
>> xlabel('Discrete time, 4*nT')
>> ylabel('Decimated output signal level')
>>
sh
a
Commands
AL
Graphs -
Part a 1000 samples
TA
AL
)-
ng
si
es
ro
c
lP
na
ig
lS
ita
ig
(D
Zooming
Ak
sh
a
ns
h
ns
h
sh
a
Ak
)ng
si
es
ro
c
lP
na
AL
TA
(D
ig
ita
lS
ig
Part b Decimation (1000/10.e. 100 samples)
ns
h
sh
a
Ak
)ng
si
es
ro
c
lP
na
AL
TA
(D
ig
ita
lS
ig
Part c- Interpolation (250*10 i.e. 2500 samples)
TA
AL
)-
ng
si
es
ro
c
lP
na
ig
lS
ita
ig
(D
Zooming
Ak
sh
a
ns
h
ns
h
sh
a
Ak
)ng
si
es
ro
c
lP
na
AL
TA
(D
ig
ita
lS
ig
Combined graphSeeing the change in the number of samples in one cycle.
ns
h
sh
a
Ak
)ng
si
es
ro
c
lP
na
(D
>> Fs=5000;
>> A=2;
>> B=1;
ig
Change 1
ita
lS
ig
OBSERVATIONS:
100
>> f2=500;
AL
TA
>> f1=
>> t=0:1/Fs:1;
>> x=A*cos(2*pi*f1*t)+B*cos(2*pi*f2*t);
>> subplot(3,1,1)
>> stem(x(1:1000))
>> xlabel('Discrete time, nT')
>> ylabel('Input signal level')
>>
AL
TA
(D
ig
ita
lS
ig
na
lP
ro
c
es
si
ng
)-
Ak
sh
a
ns
h
>>
>> %Now, decimating for part b.
>> y=decimate(x,10);
>> subplot(3,1,2)
>> stem(y(1:100))
>> xlabel('Discrete time, nT/10')
>> ylabel('Decimated output signal level')
>>
>> %Now, interpolating for part c.
>> y1=interp(y,4);
>> subplot(3,1,3)
>> stem(y1(1:400))
>> xlabel('Discrete time, 4*nT')
>> ylabel('Decimated output signal level')
>>
Change 2
Fs=5000;
>> A=2;
>> B=1;
>> f1=50;
>> f2=100;
>> t=0:1/Fs:1;
AL
TA
(D
ig
ita
lS
ig
na
lP
ro
c
es
si
ng
)-
Ak
sh
a
ns
h
sin
>> x=A*cos(2*pi*f1*t)+B*
(2*pi*f2*t);
>> subplot(3,1,1)
>> stem(x(1:1000))
>> xlabel('Discrete time, nT')
>> ylabel('Input signal level')
>>
>>
>> %Now, decimating for part b.
>> y=decimate(x,10);
>> subplot(3,1,2)
>> stem(y(1:100))
>> xlabel('Discrete time, nT/10')
>> ylabel('Decimated output signal level')
>>
>> %Now, interpolating for part c.
>> y1=interp(y,4);
>> subplot(3,1,3)
>> stem(y1(1:400))
>> xlabel('Discrete time, 4*nT')
>> ylabel('Decimated output signal level')
ns
h
sh
a
Ak
)ng
si
es
ro
c
lP
na
ig
Change 3
ita
ig
50;
>> B=-10;
>> A=
lS
Fs=5000;
AL
TA
(D
>> f1=50;
>> f2=100;
>> t=0:1/Fs:1;
>> x=A*cos(2*pi*f1*t)+B*cos(2*pi*f2*t);
>> subplot(3,1,1)
>> stem(x(1:1000))
>> xlabel('Discrete time, nT')
>> ylabel('Input signal level')
>>
>>
>> %Now, decimating for part b.
>> y=decimate(x,10);
>> subplot(3,1,2)
>> stem(y(1:100))
(D
ig
ita
lS
ig
na
lP
ro
c
es
si
ng
)-
Ak
sh
a
ns
h
>> xlabel('Discrete time, nT/10')
>> ylabel('Decimated output signal level')
>>
>> %Now, interpolating for part c.
>> y1=interp(y,4);
>> subplot(3,1,3)
>> stem(y1(1:400))
>> xlabel('Discrete time, 4*nT')
>> ylabel('Decimated output signal level')
>>
TA
Change 4
AL
Idea No. of samples taken should always be less than the sampling frequency that we take.
200
Fs=
;
>> A=2;
>> B=1;
>> f1=50;
>> f2=100;
>> t=0:1/Fs:1;
>> x=A*cos(2*pi*f1*t)+B*cos(2*pi*f2*t);
>> subplot(3,1,1)
100
sh
a
ns
h
>> stem(x(1:
))
>> xlabel('Discrete time, nT')
>> ylabel('Input signal level')
>>
>>
>> %Now, decimating for part b.
>> y=decimate(x,10);
>> subplot(3,1,2)
Ak
10
si
ng
)-
>> stem(y(1:
))
>> xlabel('Discrete time, nT/10')
>> ylabel('Decimated output signal level')
>>
>> %Now, interpolating for part c.
>> y1=interp(y,4);
>> subplot(3,1,3)
AL
TA
(D
ig
ita
lS
ig
na
lP
ro
c
es
40
>> stem(y1(1:
))
>> xlabel('Discrete time, 4*nT')
>> ylabel('Decimated output signal level')
>>
ns
h
Change 5
ig
ita
lS
ig
na
>> xlabel('Discrete time, nT/ ')
>> ylabel('Decimated output signal level')
>>
>> %Now, interpolating for part c.
AL
TA
(D
>> xlabel('Discrete time, *nT')
>> ylabel('Decimated output signal level')
Ak
)ng
si
lP
>> y=decimate(x, );
>> subplot(3,1,2)
>> stem(y(1:100))
>> y1=interp(y, );
>> subplot(3,1,3)
>> stem(y1(1:400))
es
ro
c
sh
a
>> Fs=5000;
>> A=2;
>> B=1;
>> f1=50;
>> f2=100;
>> t=0:1/Fs:1;
>> x=A*cos(2*pi*f1*t)+B*cos(2*pi*f2*t);
>> subplot(3,1,1)
>> stem(x(1:1000))
>> xlabel('Discrete time, nT')
>> ylabel('Input signal level')
>>
>>
>> %Now, decimating for part b.
ns
h
sh
a
Ak
)ng
si
es
ro
c
lP
na
lS
ita
(D
5+
ig
Fs=5000;
>> A=2;
>> B=1;
>> f1=50;
>> f2=100;
>> t=0:1/Fs:1;
ig
Change 6
AL
TA
>> x=
A*cos(2*pi*f1*t)+B*cos(2*pi*f2*t);
>> subplot(3,1,1)
>> stem(x(1:1000))
>> xlabel('Discrete time, nT')
>> ylabel('Input signal level')
>>
>>
>> %Now, decimating for part b.
>> y=decimate(x,10);
>> subplot(3,1,2)
>> stem(y(1:100))
>> xlabel('Discrete time, nT/10')
AL
TA
(D
ig
ita
lS
ig
na
lP
ro
c
es
si
ng
)-
Ak
sh
a
ns
h
>> ylabel('Decimated output signal level')
>>
>> %Now, interpolating for part c.
>> y1=interp(y,4);
>> subplot(3,1,3)
>> stem(y1(1:400))
>> xlabel('Discrete time, 4*nT')
>> ylabel('Decimated output signal level')
TA
AL
)-
ng
si
es
ro
c
lP
na
ig
lS
ita
ig
(D
Ak
sh
a
ns
h
