MATLAB Programs

Chapter 16
16.1 INTRODUCTION
MATLAB stands for MATrix LABoratory. It is a technical computing environment for high performance numeric computation and visualisation. It integrates numerical analysis, matrix computation, signal processing and graphics in an easy-to-use environment, where problems and solutions are expressed just as they are written mathematically, without traditional programming. MATLAB allows us to express the entire algorithm in a few dozen lines, to compute the solution with great accuracy in a few minutes on a computer, and to readily manipulate a three-dimensional display of the result in colour. MATLAB is an interactive system whose basic data element is a matrix that does not require dimensioning. It enables us to solve many numerical problems in a fraction of the time that it would take to write a program and execute in a language such as FORTRAN, BASIC, or C. It also features a family of application specic solutions, called toolboxes. Areas in which toolboxes are available include signal processing, image processing, control systems design, dynamic systems simulation, systems identication, neural networks, wavelength communication and others. It can handle linear, non-linear, continuous-time, discrete-time, multivariable and multirate systems. This chapter gives simple programs to solve specic problems that are included in the previous chapters. All these MATLAB programs have been tested under version 7.1 of MATLAB and version 6.12 of the signal processing toolbox.

16.2

REPRESENTATION OF BASIC SIGNALS

MATLAB programs for the generation of unit impulse, unit step, ramp, exponential, sinusoidal and cosine sequences are as follows. % Program for the generation of unit impulse signal clc;clear all;close all; t522:1:2; y5[zeros(1,2),ones(1,1),zeros(1,2)];subplot(2,2,1);stem(t,y);

816

Digital Signal Processing

ylabel(Amplitude --.); xlabel((a) n --.); % Program for the generation of unit step sequence [u(n)2 u(n 2 N] n5input(enter the N value); t50:1:n21; y15ones(1,n);subplot(2,2,2); stem(t,y1);ylabel(Amplitude --.); xlabel((b) n --.); % Program for the generation of ramp sequence n15input(enter the length of ramp sequence); t50:n1; subplot(2,2,3);stem(t,t);ylabel(Amplitude --.); xlabel((c) n --.); % Program for the generation of exponential sequence n25input(enter the length of exponential sequence); t50:n2; a5input(Enter the a value); y25exp(a*t);subplot(2,2,4); stem(t,y2);ylabel(Amplitude --.); xlabel((d) n --.); % Program for the generation of sine sequence t50:.01:pi; y5sin(2*pi*t);gure(2); subplot(2,1,1);plot(t,y);ylabel(Amplitude --.); xlabel((a) n --.); % Program for the generation of cosine sequence t50:.01:pi; y5cos(2*pi*t); subplot(2,1,2);plot(t,y);ylabel(Amplitude --.); xlabel((b) n --.); As an example, enter the N value 7 enter the length of ramp sequence 7 enter the length of exponential sequence 7 enter the a value 1 Using the above MATLAB programs, we can obtain the waveforms of the unit impulse signal, unit step signal, ramp signal, exponential signal, sine wave signal and cosine wave signal as shown in Fig. 16.1.

MATLAB Programs

817

1 0.8

1 0.8

Amplitude

0.4 0.2 0 2 1 (a) 7 6 5 4 3 2 1 0 0 2 n (c) 1 0.5 0 4 6 8 0 n 1 2

Amplitude

0.6

0.6 0.4 0.2 0 0 2 (b) 1 0.8 4 n 6

Amplitude

Amplitude

0.6 0.4 0.2 0 0 2 4 n (d) 6 8

Amplitude

0.5 1

0.5

1.5 (e)

2 n

2.5

3.5

1 0.5 0 0.5 1

Amplitude

0.5

1.5 (f)

2 n

2.5

3.5

Fig. 16.1 Representation of Basic Signals (a) Unit Impulse Signal (b) Unit-step Signal (c) Ramp Signal (d) Exponential Signal (e) Sinewave Signal ( f )Cosine Wave Signal

818

Digital Signal Processing

16.3

DISCRETE CONVOLUTION

16.3.1 Linear Convolution Algorithm 1. Get two signals x(m)and h(p)in matrix form 2. The convolved signal is denoted as y(n) 3. y(n)is given by the formula 4. Stop y(n)5
k =

[x(k ) h(n k )] where n50 to m1p21

% Program for linear convolution of the sequence x5[1, 2] and h5[1, 2, 4] clc; clear all; close all; x5input(enter the 1st sequence); h5input(enter the 2nd sequence); y5conv(x,h); gure;subplot(3,1,1); stem(x);ylabel(Amplitude --.); xlabel((a) n --.); subplot(3,1,2); stem(h);ylabel(Amplitude --.); xlabel((b) n --.); subplot(3,1,3); stem(y);ylabel(Amplitude --.); xlabel((c) n --.); disp(The resultant signal is);y As an example, enter the 1st sequence [1 2] enter the 2nd sequence [1 2 4] The resultant signal is y51 4 8 8 Figure 16.2 shows the discrete input signals x(n)and h(n)and the convolved output signal y(n).
2 1.5

Amplitude

1 0.5 0 1 1.1 1.2 1.3 1.4 1.5 (a) 4 3 1.6 1.7 1.8 n 1.9 2

Fig. 16.2 (Contd.)

Amplitude

2 1 0 1 1.2 1.4 1.6 1.8 2 (b) 2.2 2.4 2.6 n 2.8 3

Amplitude Amplitude

2 2 1.5 1.5 1 1 0.5 0.5 0 01 1 1.1 1.1 1.2 1.2 1.3 1.3 1.4 1.4 1.5 1.5 (a) (a) 1.6 1.6 1.7 1.7
MATLAB Programs

1.8 1.8 n n

1.9 1.9

2 2

819

Amplitude Amplitude

4 4 3 3 2 2 1 1 0 01 1 1.2 1.2 1.4 1.4 1.6 1.6 1.8 1.8 2 2 (b) (b) 2.2 2.2 2.4 2.4 2.6 2.6 n n 2.8 2.8 3 3

Amplitude Amplitude

8 8 6 6 4 4 2 2 0 0 1 1 1.5 1.5 2 2 2.5 2.5 (c) (c) 3 3 n n 3.5 3.5 4 4

Fig. 16.2 Discrete Linear Convolution

16.3.2 Circular Convolution % Program for Computing Circular Convolution clc; clear; a = input(enter the sequence x(n) = ); b = input(enter the sequence h(n) = ); n1=length(a); n2=length(b); N=max(n1,n2); x = [a zeros(1,(N-n1))]; for i = 1:N k = i; for j = 1:n2 H(i,j)=x(k)* b(j); k = k-1; if (k == 0) k = N; end end end y=zeros(1,N); M=H; for j = 1:N for i = 1:n2 y(j)=M(i,j)+y(j); end end disp(The output sequence is y(n)= ); disp(y);

820

Digital Signal Processing

stem(y); title(Circular Convolution); xlabel(n); ylabel(y(n)); As an Example, enter the sequence x(n) = [1 2 4] enter the sequence h(n) = [1 2] The output sequence is y(n)= 9 4 8 % Program for Computing Circular Convolution with zero padding clc; close all; clear all; g5input(enter the rst sequence); h5input(enter the 2nd sequence); N15length(g); N25length(h); N5max(N1,N2); N35N12N2; %Loop for getting equal length sequence if(N350) h5[h,zeros(1,N3)]; else g5[g,zeros(1,2N3)]; end %computation of circular convolved sequence for n51:N, y(n)50; for i51:N, j5n2i11; if(j550) j5N1j; end y(n)5y(n)1g(i)*h(j); end end disp(The resultant signal is);y As an example, enter the rst sequence [1 2 4] enter the 2nd sequence [1 2] The resultant signal is y51 4 8 8 16.3.3 Overlap Save Method and Overlap Add method % Program for computing Block Convolution using Overlap Save Method Overlap Save Method x=input(Enter the sequence x(n) = );

MATLAB Programs

821

h=input(Enter the sequence h(n) = ); n1=length(x); n2=length(h); N=n1+n2-1; h1=[h zeros(1,N-n1)]; n3=length(h1); y=zeros(1,N); x1=[zeros(1,n3-n2) x zeros(1,n3)]; H=fft(h1); for i=1:n2:N y1=x1(i:i+(2*(n3-n2))); y2=fft(y1); y3=y2.*H; y4=round(ifft(y3)); y(i:(i+n3-n2))=y4(n2:n3); end disp(The output sequence y(n)=); disp(y(1:N)); stem(y(1:N)); title(Overlap Save Method); xlabel(n); ylabel(y(n)); Enter the sequence x(n) = [1 2 -1 2 3 -2 -3 -1 1 1 2 -1] Enter the sequence h(n) = [1 2 3 -1] The output sequence y(n) = 1 4 6 5 2 11 0 -16 -8 3 8 5 3 -5 1 %Program for computing Block Convolution using Overlap Add Method x=input(Enter the sequence x(n) = ); h=input(Enter the sequence h(n) = ); n1=length(x); n2=length(h); N=n1+n2-1; y=zeros(1,N); h1=[h zeros(1,n2-1)]; n3=length(h1); y=zeros(1,N+n3-n2); H=fft(h1); for i=1:n2:n1 if i<=(n1+n2-1) x1=[x(i:i+n3-n2) zeros(1,n3-n2)]; else x1=[x(i:n1) zeros(1,n3-n2)]; end x2=fft(x1); x3=x2.*H; x4=round(ifft(x3)); if (i==1)

822

Digital Signal Processing

y(1:n3)=x4(1:n3); else y(i:i+n3-1)=y(i:i+n3-1)+x4(1:n3); end end disp(The output sequence y(n)=); disp(y(1:N)); stem((y(1:N)); title(Overlap Add Method); xlabel(n); ylabel(y(n)); As an Example, Enter the sequence x(n) = [1 2 -1 2 3 -2 -3 -1 1 1 2 -1] Enter the sequence h(n) = [1 2 3 -1] The output sequence y(n) = 1 4 6 5 2 11 0 -16 -8 3 8 5 3 -5 1

16.4
Algorithm

DISCRETE CORRELATION

16.4.1 Crosscorrelation

1. Get two signals x(m)and h(p)in matrix form 2. The correlated signal is denoted as y(n) 3. y(n)is given by the formula y(n)5
k =

[x(k ) h(k n)]

where n52 [max (m, p)2 1] to [max (m, p)2 1] 4. Stop % Program for computing cross-correlation of the sequences x5[1, 2, 3, 4] and h5[4, 3, 2, 1] clc; clear all; close all; x5input(enter the 1st sequence); h5input(enter the 2nd sequence); y5xcorr(x,h); gure;subplot(3,1,1); stem(x);ylabel(Amplitude --.); xlabel((a) n --.); subplot(3,1,2); stem(h);ylabel(Amplitude --.); xlabel((b) n --.); subplot(3,1,3); stem(iplr(y));ylabel(Amplitude --.);

MATLAB Programs

823

xlabel((c) n --.); disp(The resultant signal is);iplr(y) As an example, enter the 1st sequence [1 2 3 4] enter the 2nd sequence [4 3 2 1] The resultant signal is y51.0000 4.0000 10.0000 20.0000 25.0000 24.0000 16.0000 Figure 16.3 shows the discrete input signals x(n)and h(n)and the cross-correlated output signal y(n).
4 3

Amplitude

2 1 0 1 1.5 2 2.5 (a) 4 3 3 n 3.5 4

Amplitude

2 1 0 1 1.5 2 2.5 (b) 30 3 n 3.5 4

Amplitude

20 10 0 1 2 3 4 (c) 5 n 6 7

Fig. 16.3 Discrete Cross-correlation

16.4.2 Autocorrelation Algorithm 1. Get the signal x(n)of length N in matrix form 2. The correlated signal is denoted as y(n) 3. y(n)is given by the formula y(n)5
k =

[x(k ) x(k n)]

where n52(N 2 1) to (N 2 1)

824

Digital Signal Processing

% Program for computing autocorrelation function x5input(enter the sequence); y5xcorr(x,x); gure;subplot(2,1,1); stem(x);ylabel(Amplitude --.); xlabel((a) n --.); subplot(2,1,2); stem(iplr(y));ylabel(Amplitude --.); xlabel((a) n --.); disp(The resultant signal is);iplr(y) As an example, enter the sequence [1 2 3 4] The resultant signal is y54 11 20 30 20 11 4 Figure 16.4 shows the discrete input signal x(n)and its auto-correlated output signal y(n).
4 3

Amplitude

2 1 0 1 1.5 2 2.5 (a) 30 25 20 15 10 5 0 3 3.5 n 4

( )

Amplitude

4 (b) y (n)

6 n

Fig. 16.4 Discrete Auto-correlation

16.5

STABILITY TEST

% Program for stability test clc;clear all;close all; b5input(enter the denominator coefcients of the lter); k5poly2rc(b); knew5iplr(k); s5all(abs(knew)1); if(s55 1) disp(Stable system);

MATLAB Programs

825

else disp(Non-stable system); end As an example, enter the denominator coefcients of the lter [1 21 .5] Stable system

16.6

SAMPLING THEOREM

The sampling theorem can be understood well with the following example. Example 16.1 Frequency analysis of the amplitude modulated discrete-time signal x(n)5cos 2 pf1n1cos 2pf2n 5 1 where f1 = and f 2 = modulates the amplitude-modulated signal is 128 128 xc(n)5cos 2p fc n where fc550/128. The resulting amplitude-modulated signal is xam(n)5x(n) cos 2p fc n Using MATLAB program, (a) sketch the signals x(n), xc(n) and xam(n), 0#n#255 (b) compute and sketch the 128-point DFT of the signal xam(n), 0#n#127 (c) compute and sketch the 128-point DFT of the signal xam(n), 0#n#99 Solution % Program Solution for Section (a) clc;close all;clear all; f151/128;f255/128;n50:255;fc550/128; x5cos(2*pi*f1*n)1cos(2*pi*f2*n); xa5cos(2*pi*fc*n); xamp5x.*xa; subplot(2,2,1);plot(n,x);title(x(n)); xlabel(n --.);ylabel(amplitude); subplot(2,2,2);plot(n,xc);title(xa(n)); xlabel(n --.);ylabel(amplitude); subplot(2,2,3);plot(n,xamp); xlabel(n --.);ylabel(amplitude); %128 point DFT computation2solution for Section (b) n50:127;gure;n15128; f151/128;f255/128;fc550/128; x5cos(2*pi*f1*n)1cos(2*pi*f2*n); xc5cos(2*pi*fc*n); xa5cos(2*pi*fc*n); (Contd.)

826

Digital Signal Processing

Amplitude

100

(i)

200

300

Fig. 16.5(a) (i) Modulating Signal x (n)

1

0.5

Amplitude

0.5

1 0 100 (ii) 200 n 300

Fig. 16.5(a) (ii) Carrier Signal and

2

Amplitude

100 (iii)

200 n

300

Fig. 16.5(a) (iii) Amplitude Modulated Signal

(Contd.)

MATLAB Programs

827

25

20

15

10

Amplitude

10

20

40

60

80

100 n

120

140

Fig. 16.5(b) 128-point DFT of the Signal xam (n), 0#n#127

35

30

25

20

Amplitude

15

10

20

40

60

80

100

120

140

Fig. 16.5(c) 128-point DFT of the Signal xam (n), 0#n#99

828

Digital Signal Processing

xamp5x.*xa;xam5fft(xamp,n1); stem(n,xam);title(xamp(n));xlabel(n --.); ylabel(amplitude); %128 point DFT computation2solution for Section (c) n50:99;gure;n250:n121; f151/128;f255/128;fc550/128; x5cos(2*pi*f1*n)1cos(2*pi*f2*n); xc5cos(2*pi*fc*n); xa5cos(2*pi*fc*n); xamp5x.*xa; for i51:100, xamp1(i)5xamp(i); end xam5fft(xamp1,n1); stem(n2,xam);title(xamp(n));xlabel(n --.);ylabel(amplitude); (a)Modulated signal x(n), carrier signal xa(n) and amplitude modulated signal xam(n) are shown in Fig. 16.5(a). Fig. 16.5 (b) shows the 128-point DFT of the signal xam(n) for 0#n#127 and Fig. 16.5 (c) shows the 128-point DFT of the signal xam(n), 0#n#99.

16.7
Algorithm

FAST FOURIER TRANSFORM

1. Get the signal x(n)of length N in matrix form 2. Get the N value 3. The transformed signal is denoted as x( k ) = x( n )e
n=0 N 1 j 2p nk N

for 0 k N 1

\ \% Program for computing discrete Fourier transform clc;close all;clear all; x5input(enter the sequence); n5input(enter the length of fft); X(k)5fft(x,n); stem(y);ylabel(Imaginary axis --.); xlabel(Real axis --.); X(k) As an example, enter the sequence [0 1 2 3 4 5 6 7] enter the length of fft 8 X(k)5 Columns 1 through 4 28.0000 24.000019.6569i 24.0000 14.0000i 24.0000 1 1.6569i Columns 5 through 8 24.0000 24.0000 21.6569i 24.0000 24.0000i 24.0000 29.6569i

MATLAB Programs

829

The eight-point decimation-in-time fast Fourier transform of the sequence x(n)is computed using MATLAB program and the resultant output is plotted in Fig. 16.6.
10 8 6 4 2

Imaginary axis

0 2 4 6 8 10

10 Real axis

15

20

25

30

Fig. 16.6 Fast Fourier Transform

16.8

BUTTERWORTH ANALOG FILTERS

16.8.1 Low-pass Filter Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter using Eq. 8.46 Find the lter coefcients Draw the magnitude and phase responses.

% Program for the design of Butterworth analog low pass lter clc; close all;clear format long rp5input(enter rs5input(enter wp5input(enter ws5input(enter fs5input(enter all; the the the the the passband stopband passband stopband sampling ripple); ripple); freq); freq); freq);

830

Digital Signal Processing

w152*wp/fs;w252*ws/fs; [n,wn]5buttord(w1,w2,rp,rs,s); [z,p,k]5butter(n,wn); [b,a]5zp2tf(z,p,k); [b,a]5butter(n,wn,s); w50:.01:pi; [h,om]5freqs(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.); As an example, enter the passband ripple 0.15 enter the stopband ripple 60 enter the passband freq 1500 enter the stopband freq 3000 enter the stopband freq 7000 The amplitude and phase responses of the Butterworth low-pass analog lter are shown in Fig. 16.7.
50 50 0 0 50 50 100 100 150 150 200 200 250 250 0 0

Gain Gain in in dB dB

0.1 0.1

0.2 0.2

0.3 0.3

0.4 0.4 (a) (a)

0.5 0.5

0.6 0.7 0.8 0.9 0.6 0.7 0.8 0.9 Normalised frequency Normalised frequency

1 1

4 4 2 2 0 0 2 2 4 4 0 0 0.1 0.1 0.2 0.2 0.3 0.3 0.4 0.4 (b) (b) 0.5 0.5 0.6 0.7 0.8 0.9 0.6 0.7 0.8 0.9 Normalised frequency Normalised frequency 1 1

Phase Phase in in radians radians

Fig. 16.7 Butterworth Low-pass Analog Filter (a) Amplitude Response and (b) Phase Response

MATLAB Programs

831

16.8.2 High-pass Filter Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter using Eq. 8.46 Find the lter coefcients Draw the magnitude and phase responses.

% Program for the design of Butterworth analog highpass lter clc; close all;clear all; format long rp5input(enter the passband ripple); rs5input(enter the stopband ripple); wp5input(enter the passband freq); ws5input(enter the stopband freq); fs5input(enter the sampling freq); w152*wp/fs;w252*ws/fs; [n,wn]5buttord(w1,w2,rp,rs,s); [b,a]5butter(n,wn,high,s); w50:.01:pi; [h,om]5freqs(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.); As an example, enter the passband enter the stopband enter the passband enter the stopband enter the sampling ripple ripple freq freq freq 0.2 40 2000 3500 8000

The amplitude and phase responses of Butterworth high-pass analog lter are shown in Fig. 16.8.

832

Digital Signal Processing

100 100 0

Gain Gain in dB in dB

0 100 100 200 200 300 300 400 400 0 0 0.1 0.1 0.2 0.2 0.3 0.3 0.4 0.4 (a) (a) 4 4 2 0.5 0.5 0.6 0.7 0.8 0.9 1 1 0.6 0.7 frequency 0.8 0.9 Normalised Normalised frequency

Phase Phase in radians in radians

2 0 0 2 2 4

4 0 0

0.1 0.1

0.2 0.2

0.3 0.3

0.4 0.4 (b)

0.5 0.5

0.6

0.7

0.8

0.9

1 1

0.6 0.7 frequency 0.8 0.9 Normalised Normalised frequency

Fig. 16.8 Butterworth High-pass Analog (b) Filter (a) Amplitude Response and (b) Phase Response

16.8.3 Bandpass Filter Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter using Eq. 8.46 Find the lter coefcients Draw the magnitude and phase responses.

% Program for the design of Butterworth analog Bandpass lter clc; close all;clear all; format long rp5input(enter the passband rs5input(enter the stopband wp5input(enter the passband ws5input(enter the stopband fs5input(enter the sampling w152*wp/fs;w252*ws/fs;

ripple...); ripple...); freq...); freq...); freq...);

MATLAB Programs

833

[n]5buttord(w1,w2,rp,rs); wn5[w1 w2]; [b,a]5butter(n,wn,bandpass,s); w50:.01:pi; [h,om]5freqs(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.); As an example, enter the passband ripple... 0.36 enter the stopband ripple... 36 enter the passband freq... 1500 enter the stopband freq... 2000 enter the sampling freq... 6000 The amplitude and phase responses of Butterworth bandpass analog lter are shown in Fig. 16.9.
200 0 200

Gain in dB

400 600 800 1000 0 0.1 0.2 0.3 0.4 (a) 4 2 0.5 0.6 0.7 0.8 0.9 1

Normalised frequency

Phase in radians

0 2 4

0.1

0.2

0.3

0.4 (b)

0.5

0.6

0.7

0.8

0.9

Normalised frequency

Fig. 16.9 Butterworth Bandpass Analog Filter (a) Amplitude Response and (b) Phase Response

834

Digital Signal Processing

16.8.4 Bandstop Filter Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter using Eq. 8.46 Find the lter coefcients Draw the magnitude and phase responses.

% Program for the design of Butterworth analog Bandstop lter clc; close all;clear all; format long rp5input(enter the passband ripple...); rs5input(enter the stopband ripple...); wp5input(enter the passband freq...); ws5input(enter the stopband freq...); fs5input(enter the sampling freq...); w152*wp/fs;w252*ws/fs; [n]5buttord(w1,w2,rp,rs,s); wn5[w1 w2]; [b,a]5butter(n,wn,stop,s); w50:.01:pi; [h,om]5freqs(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.); As an example, enter the passband enter the stopband enter the passband enter the stopband enter the sampling ripple... ripple... freq... freq... freq... 0.28 28 1000 1400 5000

The amplitude and phase responses of Butterworth bandstop analog lter are shown in Fig. 16.10.

MATLAB Programs

835

50 50 0 0

Gain Gain inin dB dB

50 50 100 100 150 150 200 200 0 0 0.1 0.1 0.2 0.2 0.3 0.3 0.4 0.4 (a) (a) 4 4 0.5 0.5 0.6 0.7 0.8 0.9 0.6 0.7 0.8 0.9 Normalised frequency Normalised frequency 1 1

Phase Phase inin radians radians

2 2 0 0 2 2 4 4 0 0 0.1 0.1 0.2 0.2 0.3 0.3 0.4 0.4 (b) (b) 0.5 0.5 0.6 0.7 0.8 0.9 0.6 0.7 0.8 0.9 Normalised frequency Normalised frequency 1 1

Fig. 16.10 Butterworth Bandstop Analog Filter (a) Amplitude Response and (b) Phase Response

16.9

CHEBYSHEV TYPE-1 ANALOG FILTERS

16.9.1 Low-pass Filter Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter using Eq. 8.57 Find the lter coefcients Draw the magnitude and phase responses.

% Program for the design of Chebyshev Type-1 low-pass lter clc; close all;clear format long rp5input(enter rs5input(enter wp5input(enter ws5input(enter fs5input(enter all; the the the the the passband stopband passband stopband sampling ripple...); ripple...); freq...); freq...); freq...);

836

Digital Signal Processing

w152*wp/fs;w252*ws/fs; [n,wn]5cheb1ord(w1,w2,rp,rs,s); [b,a]5cheby1(n,rp,wn,s); w50:.01:pi; [h,om]5freqs(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.); As an example, enter the passband ripple... 0.23 enter the stopband ripple... 47 enter the passband freq... 1300 enter the stopband freq... 1550 enter the sampling freq... 7800 The amplitude and phase responses of Chebyshev type - 1 low-pass analog lter are shown in Fig. 16.11.
0 20

Gain in dB

40 60 80 100 0 0.1 0.2 0.3 0.4 (a) 4 2 0.5 0.6 0.7 0.8 0.9 1

Normalised frequency

Phase in radians

0 2 4

0.1

0.2

0.3

0.4 (b)

0.5

0.6

0.7

0.8

0.9

Normalised frequency

Fig. 16.11 Chebyshev Type-I Low-pass Analog Filter (a) Amplitude Response and (b) Phase Response

MATLAB Programs

837

16.9.2 High-pass Filter Algorithm 1. Get the passband and stopband ripples 2. Get the passband and stopband edge frequencies 3. Get the sampling frequency 4. Calculate the order of the lter using Eq. 8.57 5. Find the lter coefcients 6. Draw the magnitude and phase responses. %Program for the design of Chebyshev Type-1 high-pass lter clc; close all;clear all; format long rp5input(enter the passband ripple...); rs5input(enter the stopband ripple...); wp5input(enter the passband freq...); ws5input(enter the stopband freq...); fs5input(enter the sampling freq...); w152*wp/fs;w252*ws/fs; [n,wn]5cheb1ord(w1,w2,rp,rs,s); [b,a]5cheby1(n,rp,wn,high,s); w50:.01:pi; [h,om]5freqs(b,a,w); m520*log10(abs(h)); an5angle(h); 0 subplot(2,1,1);plot(om/pi,m);
0 50

Gain Gain in dB in dB

50 100 100 150 150 200 200

0 0

0.1 0.1

0.2 0.2

0.3 0.3

0.4 0.4 (a) (a)

0.5 0.5

0.6

0.7

0.8

0.9 0.9

1 1

0.6 0.7 frequency 0.8 Normalised Normalised frequency

4 4 2

Phase in radians Phase in radians

2 0 0 2 2 4 4

0 0

0.1 0.1

0.2 0.2

0.3 0.3

0.4 0.4 (b)

0.5 0.5

0.6

0.7

0.8

0.9 0.9

1 1

0.6 0.7 frequency 0.8 Normalised

Normalised frequency Fig. 16.12 Chebyshev Type - 1 High-pass (b) Analog Filter (a) Amplitude Response and (b) Phase Response

838

Digital Signal Processing

ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.); As an example, enter the passband ripple... 0.29 enter the stopband ripple... 29 enter the passband freq... 900 enter the stopband freq... 1300 enter the sampling freq... 7500 The amplitude and phase responses of Chebyshev type - 1 high-pass analog lter are shown in Fig. 16.12. 16.9.3 Bandpass Filter Algorithm 1. Get the passband and stopband ripples 2. Get the passband and stopband edge frequencies 3. Get the sampling frequency 4. Calculate the order of the lter using Eq. 8.57 5. Find the lter coefcients 6. Draw the magnitude and phase responses. % Program for the design of Chebyshev Type-1 Bandpass lter clc; close all;clear all; format long rp5input(enter the passband ripple...); rs5input(enter the stopband ripple...); wp5input(enter the passband freq...); ws5input(enter the stopband freq...); fs5input(enter the sampling freq...); w152*wp/fs;w252*ws/fs; [n]5cheb1ord(w1,w2,rp,rs,s); wn5[w1 w2]; [b,a]5cheby1(n,rp,wn,bandpass,s); w50:.01:pi; [h,om]5freqs(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.); As an example, enter the passband ripple... 0.3 enter the stopband ripple... 40 enter the passband freq... 1400

MATLAB Programs

839

enter the stopband freq... enter the sampling freq...

2000 5000

The amplitude and phase responses of Chebyshev type - 1 bandpass analog lter are shown in Fig. 16.13.
0 0 100

Gain Gain in dB in dB

100 200 200 300 300 400 400

0 0

0.1 0.1

0.2 0.2

0.3 0.3

0.4 0.4 (a) (a)

0.5 0.5

0.6

0.7

0.8

0.9

1 1

0.6 Normalised 0.7 0.8 0.9 frequency Normalised frequency

3 3 2 2 1

Phase in radians Phase in radians

1 0 0 1 1 2 2 3 3 0 0 0.1 0.1 0.2 0.2 0.3 0.3 0.4 0.4 (b) 0.5 0.5 0.6 0.7 0.8 0.9 1 1

0.6 0.7 0.8 0.9 Normalised frequency Normalised frequency

(b) Bandpass Analog Filter Fig. 16.13 Chebyshev Type-1 (a) Amplitude Response and (b) Phase Response

16.9.4 Bandstop Filter Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequency Get the sampling frequency Calculate the order of the lter using Eq. 8.57 Find the lter coefcients Draw the magnitude and phase responses.

% Program for the design of Chebyshev Type-1 Bandstop lter clc; close all;clear format long rp5input(enter rs5input(enter wp5input(enter ws5input(enter all; the the the the passband stopband passband stopband ripple...); ripple...); freq...); freq...);

840

Digital Signal Processing

fs5input(enter the sampling freq...); w152*wp/fs;w252*ws/fs; [n]5cheb1ord(w1,w2,rp,rs,s); wn5[w1 w2]; [b,a]5cheby1(n,rp,wn,stop,s); w50:.01:pi; [h,om]5freqs(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.); As an example, enter the passband ripple... 0.15 enter the stopband ripple... 30 enter the passband freq... 2000 enter the stopband freq... 2400 enter the sampling freq... 7000 The amplitude and phase responses of Chebyshev type - 1 bandstop analog lter are shown in Fig. 16.14.
0 0 50

Gain Gain in dB in dB

50 100 100 150 150 200 200 250 250 0 0 0.1 0.1 0.2 0.2 0.3 0.3 0.4 0.4 (a) (a) 4 4 2 0.5 0.5 0.6 0.7 0.8 0.9 1 1

0.6 0.7 0.8 0.9 Normalised frequency Normalised frequency

Phase Phase in radians in radians

2 0 0 2 2 4

4 0 0

0.1 0.1

0.2 0.2

0.3 0.3

0.4 0.4 (b)

0.5 0.5

0.6

0.7

0.8

0.9

1 1

0.6 0.7 0.8 0.9 Normalised frequency Normalised frequency

(b) - 1 Bandstop Analog Filter Fig. 16.14 Chebyshev Type (a) Amplitude Response and (b) Phase Response

MATLAB Programs

841

16.10 CHEBYSHEV TYPE-2 ANALOG FILTERS

16.10.1 Low-pass Filter Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter using Eq. 8.67 Find the lter coefcients Draw the magnitude and phase responses.

% Program for the design of Chebyshev Type-2 low pass analog lter clc; close all;clear all; format long rp5input(enter the passband ripple...); rs5input(enter the stopband ripple...); wp5input(enter the passband freq...); ws5input(enter the stopband freq...); fs5input(enter the sampling freq...); w152*wp/fs;w252*ws/fs; [n,wn]5cheb2ord(w1,w2,rp,rs,s); [b,a]5cheby2(n,rs,wn,s); w50:.01:pi; [h,om]5freqs(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.); As an example, enter the passband ripple... 0.4 enter the stopband ripple... 50 enter the passband freq... 2000 enter the stopband freq... 2400 enter the sampling freq... 10000 The amplitude and phase responses of Chebyshev type - 2 low-pass analog lter are shown in Fig. 16.15.

842

Digital Signal Processing

0 20

Gain in dB

40 60 80 100 0 0.1 0.2 0.3 0.4 (a) 4 2 0.5 0.6 0.7 0.8 0.9 1

Normalised frequency

Phase in radians

0 2 4

0.1

0.2

0.3

0.4 (b)

0.5

0.6

0.7

0.8

0.9

Normalised frequency

Fig. 16.15 Chebyshev Type - 2 Low-pass Analog Filter (a) Amplitude Response and (b) Phase Response

16.10.2 High-pass Filter Algorithm 1. Get the passband and stopband ripples 2. Get the passband and stopband edge frequencies 3. Get the sampling frequency 4. Calculate the order of the lter using Eq. 8.67 5. Find the lter coefcients 6. Draw the magnitude and phase responses. % Program for the design of Chebyshev Type-2 High pass analog lter clc; close all;clear all; format long rp5input(enter the passband ripple...); rs5input(enter the stopband ripple...); wp5input(enter the passband freq...); ws5input(enter the stopband freq...); fs5input(enter the sampling freq...); w152*wp/fs;w252*ws/fs; [n,wn]5cheb2ord(w1,w2,rp,rs,s); [b,a]5cheby2(n,rs,wn,high,s); w50:.01:pi;

MATLAB Programs

843

[h,om]5freqs(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.); As an example, enter the passband ripple... 0.34 enter the stopband ripple... 34 enter the passband freq... 1400 enter the stopband freq... 1600 enter the sampling freq... 10000 The amplitude and phase responses of Chebyshev type - 2 high-pass analog lter are shown in Fig. 16.16.
0 20

Gain in dB

40 60 80

0.1

0.2

0.3

0.4 (a)

0.5

0.6

0.7

0.8

0.9

Normalised frequency

4 2 0 2 4

Phase in radians

0.1

0.2

0.3

0.4 (b)

0.5

0.6

0.7

0.8

0.9

Normalised frequency

Fig. 16.16 Chebyshev Type - 2 High-pass Analog Filter (a) Amplitude Response and (b) Phase Response

16.10.3 Bandpass Filter Algorithm 1. Get the passband and stopband ripples 2. Get the passband and stopband edge frequencies

844

Digital Signal Processing

3. 4. 5. 6.

Get the sampling frequency Calculate the order of the lter using Eq. 8.67 Find the lter coefcients Draw the magnitude and phase responses.

% Program for the design of Chebyshev Type-2 Bandpass analog lter clc; close all;clear all; format long rp5input(enter the passband ripple...); rs5input(enter the stopband ripple...); wp5input(enter the passband freq...); ws5input(enter the stopband freq...); fs5input(enter the sampling freq...); w152*wp/fs;w252*ws/fs; [n]5cheb2ord(w1,w2,rp,rs,s); wn5[w1 w2]; [b,a]5cheby2(n,rs,wn,bandpass,s); w50:.01:pi; [h,om]5freqs(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.); As an example, enter the passband ripple... 0.37 enter the stopband ripple... 37 enter the passband freq... 3000 enter the stopband freq... 4000 enter the sampling freq... 9000 The amplitude and phase responses of Chebyshev type - 2 bandpass analog lter are shown in Fig. 16.17.
20 0 20

Gain in dB

40 60 80 100 0 0.1 0.2 0.3 0.4 (a) 4 2 0.5 0.6 0.7 0.8 0.9 1

Normalised frequency

Fig. 16.17 (Contd.)

ase in radians

0 2

Gain in d

40 60 80 100 0 0.1 0.2 0.3 0.4 (a) 4 2 0.5 0.6 0.7 0.8 0.9 1

MATLAB Programs Normalised frequency

845

Phase in radians

0 2 4

0.1

0.2

0.3

0.4 (b)

0.5

0.6

0.7

0.8

0.9

Normalised frequency

Fig. 16.17 Chebyshev Type - 2 Bandstop Analog Filter (a) Amplitude Response and (b) Phase Response

16.10.4 Bandstop Filter Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter using Eq. 8.67 Find the lter coefcients Draw the magnitude and phase responses.

% Program for the design of Chebyshev Type-2 Bandstop analog lter clc; close all;clear all; format long rp5input(enter the passband ripple...); rs5input(enter the stopband ripple...); wp5input(enter the passband freq...); ws5input(enter the stopband freq...); fs5input(enter the sampling freq...); w152*wp/fs;w252*ws/fs; [n]5cheb2ord(w1,w2,rp,rs,s); wn5[w1 w2]; [b,a]5cheby2(n,rs,wn,stop,s); w50:.01:pi; [h,om]5freqs(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.);

846

Digital Signal Processing

As an example, enter the passband enter the stopband enter the passband enter the stopband enter the sampling

ripple... ripple... freq... freq... freq...

0.25 30 1300 2000 8000

The amplitude and phase responses of Chebyshev type - 2 bandstop analog lter are shown in Fig. 16.18.
40 20

Gain in dB

0 20 40 60 80 0 0.1 0.2 0.3 0.4 (a) 4 2 0.5 0.6 0.7 0.8 0.9 1

Normalised frequency

Phase in radians

0 2 4

0.1

0.2

0.3

0.4 (b)

0.5

0.6

0.7

0.8

0.9

Normalised frequency

Fig. 16.18 Chebyshev Type - 2 Bandstop Analog Filter (a) Amplitude Response and (b) Phase Response

16.11 BUTTERWORTH DIGITAL IIR FILTERS

16.11.1 Low-pass Filter Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter using Eq. 8.46 Find the lter coefcients Draw the magnitude and phase responses.

MATLAB Programs

847

% Program for the design of Butterworth low pass digital lter clc; close all;clear all; format long rp5input(enter the passband ripple); rs5input(enter the stopband ripple); wp5input(enter the passband freq); ws5input(enter the stopband freq); fs5input(enter the sampling freq); w152*wp/fs;w252*ws/fs; [n,wn]5buttord(w1,w2,rp,rs); [b,a]5butter(n,wn); w50:.01:pi; [h,om]5freqz(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.); As an example, enter the passband ripple 0.5 enter the stopband ripple 50 enter the passband freq 1200 enter the stopband freq 2400 enter the sampling freq 10000 The amplitude and phase responses of Butterworth low-pass digital lter are shown in Fig. 16.19.
100 0

Gain in dB

100 200 300 400 0 0.1 0.2 0.3 0.4 (a) 4 2 0.5 0.6 0.7 0.8 0.9 1

Normalised frequency

Fig. 16.19 (Contd.)

Phase in radians

0 2 4

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Gain

200 300 400 0 0.1 0.2 0.3 0.4 (a) 0.5 0.6 0.7 0.8 0.9 1

848

Digital Signal Processing

Normalised frequency

4 2

Phase in radians

0 2 4

0.1

0.2

0.3

0.4 (b)

0.5

0.6

0.7

0.8

0.9

Normalised frequency

Fig. 16.19 Butterworth Low-pass Digital Filter (a) Amplitude Response and (b) Phase Response

16.11.2 High-pass Filter Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter using Eq. 8.46 Find the lter coefcients Draw the magnitude and phase responses.

% Program for the design of Butterworth highpass digital lter clc; close all;clear all; format long rp5input(enter the passband ripple); rs5input(enter the stopband ripple); wp5input(enter the passband freq); ws5input(enter the stopband freq); fs5input(enter the sampling freq); w152*wp/fs;w252*ws/fs; [n,wn]5buttord(w1,w2,rp,rs); [b,a]5butter(n,wn,high); w50:.01:pi; [h,om]5freqz(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.); As an example, enter the passband ripple 0.5 enter the stopband ripple 50 enter the passband freq 1200

MATLAB Programs

849

enter the stopband freq enter the sampling freq

2400 10000

The amplitude and phase responses of Butterworth high-pass digital lter are shown in Fig. 16.20.
50 50 0 0 50 50 100 100 150 150 200 200 250 250 300 300 0 0

Gain Gain inin dB dB

0.1 0.1

0.2 0.2

0.3 0.3

0.4 0.4 (a) (a)

0.5 0.5

0.6 0.7 0.8 0.9 0.6 0.7 0.8 0.9 Normalised frequency Normalised frequency

1 1

4 4 2 2 0 0 2 2 4 4 0 0 0.1 0.1 0.2 0.2 0.3 0.3 0.4 0.4 (b) (b) 0.5 0.5 0.6 0.7 0.8 0.9 0.6 0.7 0.8 0.9 Normalised frequency Normalised frequency 1 1

Phase Phase inin radians radians

Fig. 16.20 Butterworth High-pass Digital Filter (a) Amplitude Response and (b) Phase Response

16.11.3 Band-pass Filter Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter using Eq. 8.46 Find the lter coefcients Draw the magnitude and phase responses.

% Program for the design of Butterworth Bandpass digital lter clc; close all;clear format long rp5input(enter rs5input(enter wp5input(enter all; the passband ripple); the stopband ripple); the passband freq);

850

Digital Signal Processing

ws5input(enter the stopband freq); fs5input(enter the sampling freq); w152*wp/fs;w252*ws/fs; [n]5buttord(w1,w2,rp,rs); wn5[w1 w2]; [b,a]5butter(n,wn,bandpass); w50:.01:pi; [h,om]5freqz(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.); As an example, enter the passband ripple 0.3 enter the stopband ripple 40 enter the passband freq 1500 enter the stopband freq 2000 enter the sampling freq 9000 The amplitude and phase responses of Butterworth band-pass digital lter are shown in Fig. 16.21.
0 100 200 300 400 500 600 700 0 0.1 0.2 0.3 0.4 (a) 4 2 0.5 0.6 0.7 0.8 0.9 1

Gain in dB

Normalised frequency

Phase in radians

0 2 4

0.1

0.2

0.3

0.4 (b)

0.5

0.6

0.7

0.8

0.9

Normalised frequency

Fig. 16.21 Butterworth Bandstop Digital Filter (a) Amplitude Response and (b) Phase Response

MATLAB Programs

851

16.11.4 Bandstop Filter Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter using Eq. 8.46 Find the lter coefcients Draw the magnitude and phase responses.

% Program for the design of Butterworth Band stop digital lter clc; close all;clear all; format long rp5input(enter the passband ripple); rs5input(enter the stopband ripple); wp5input(enter the passband freq); ws5input(enter the stopband freq); fs5input(enter the sampling freq); w152*wp/fs;w252*ws/fs; [n]5buttord(w1,w2,rp,rs); wn5[w1 w2]; [b,a]5butter(n,wn,stop); w50:.01:pi; [h,om]5freqz(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.); As an example, enter the passband ripple 0.4 enter the stopband ripple 46 enter the passband freq 1100 enter the stopband freq 2200 enter the sampling freq 6000 The amplitude and phase responses of the Butterworth bandstop digital lter are shown in Fig. 16.22.

852

Digital Signal Processing

100 0 100

Gain Gain in in dB dB

200 300 400 0 0.1 0.2 0.3 0.4 (a) 4 2 0.5 0.6 0.7 0.8 0.9 1

Normalised frequency

Phase Phase in in radians radians

0 2 4

0.1

0.2

0.3

0.4 (b)

0.5

0.6

0.7

0.8

0.9

Normalised frequency

Fig. 16.22 Butterworth Bandstop Digital Filter (a) Amplitude Response and (b) Phase Response

16.12 CHEBYSHEV TYPE-1 DIGITAL FILTERS

16.12.1 Low-pass Filter Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter using Eq. 8.57 Find the lter coefcients Draw the magnitude and phase responses.

% Program for the design of Chebyshev Type-1 lowpass digital lter clc; close all;clear all; format long rp5input(enter the passband ripple...); rs5input(enter the stopband ripple...); wp5input(enter the passband freq...); ws5input(enter the stopband freq...); fs5input(enter the sampling freq...);

MATLAB Programs

853

w152*wp/fs;w252*ws/fs; [n,wn]5cheb1ord(w1,w2,rp,rs); [b,a]5cheby1(n,rp,wn); w50:.01:pi; [h,om]5freqz(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.); As an example, enter the passband ripple... 0.2 enter the stopband ripple... 45 enter the passband freq... 1300 enter the stopband freq... 1500 enter the sampling freq... 10000 The amplitude and phase responses of Chebyshev type - 1 low-pass digital lter are shown in Fig. 16.23.
0 100 200

Gain in dB

300 400 500 0 0.1 0.2 0.3 0.4 (a) 4 2 0.5 0.6 0.7 0.8 0.9 1

Normalised frequency

Phase in radians

0 2 4

0.1

0.2

0.3

0.4 (b)

0.5

0.6

0.7

0.8

0.9

Normalised frequency

Fig. 16.23 Chebyshev Type - 1 Low-pass Digital Filter (a) Amplitude Response and (b) Phase Response

854

Digital Signal Processing

16.12.2 High-pass Filter Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter using Eq. 8.57 Find the lter coefcients Draw the magnitude and phase responses.

% Program for the design of Chebyshev Type-1 highpass digital filter clc; close all;clear all; format long rp5input(enter the passband ripple...); rs5input(enter the stopband ripple...); wp5input(enter the passband freq...); ws5input(enter the stopband freq...); fs5input(enter the sampling freq...); w152*wp/fs;w252*ws/fs; [n,wn]5cheb1ord(w1,w2,rp,rs); [b,a]5cheby1(n,rp,wn,high); w50:.01/pi:pi; [h,om]5freqz(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.); As an example, enter the passband enter the stopband enter the passband enter the stopband enter the sampling ripple... ripple... freq... freq... freq... 0.3 60 1500 2000 9000

The amplitude and phase responses of Chebyshev type - 1 high-pass digital lter are shown in Fig. 16.24.

MATLAB Programs

855

0 0 50 50 100 100 150 150 200 200 250 250 300 300 350 350

Gain Gain in in dB dB

0 0

0.1 0.1

0.2 0.2

0.3 0.3

0.4 0.4 (a) (a)

0.5 0.5

0.6 0.7 0.8 0.6 0.7 0.8 Normalised frequency Normalised frequency

0.9 0.9

1 1

4 4 2 2

Phase Phase in in radians radians

0 0 2 2 4 4 0 0 0.1 0.1 0.2 0.2 0.3 0.3 0.4 0.4 (b) (b) 0.5 0.5 0.6 0.7 0.8 0.6 0.7 0.8 Normalised frequency Normalised frequency 0.9 0.9 1 1

Fig. 16.24 Chebyshev Type - 1 High-pass Digital Filter (a) Amplitude Response and (b) Phase Response

16.12.3 Bandpass Filter Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter using Eq. 8.57 Find the lter coefcients Draw the magnitude and phase responses.

% Program for the design of Chebyshev Type-1 Bandpass digital lter clc; close all;clear all; format long rp5input(enter the passband rs5input(enter the stopband wp5input(enter the passband ws5input(enter the stopband fs5input(enter the sampling w152*wp/fs;w252*ws/fs;

ripple...); ripple...); freq...); freq...); freq...);

856

Digital Signal Processing

[n]5cheb1ord(w1,w2,rp,rs); wn5[w1 w2]; [b,a]5cheby1(n,rp,wn,bandpass); w50:.01:pi; [h,om]5freqz(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.); As an example, enter the passband ripple... 0.4 enter the stopband ripple... 35 enter the passband freq... 2000 enter the stopband freq... 2500 enter the sampling freq... 10000 The amplitude and phase responses of Chebyshev type - 1 bandpass digital lter are shown in Fig. 16.25.
0 100

Gain in dB

200 300 400 500 0 0.1 0.2 0.3 0.4 (a) 4 2 0.5 0.6 0.7 0.8 0.9 1

Normalised frequency

Phase in radians

0 2 4

0.1

0.2

0.3

0.4 (b)

0.5

0.6

0.7

0.8

0.9

Normalised frequency

Fig. 16.25 Chebyshev Type - 1 Bandpass Digital Filter (a) Amplitude Response and (b) Phase Response

MATLAB Programs

857

16.12.4 Bandstop Filter Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter using Eq. 8.57 Find the lter coefcients Draw the magnitude and phase responses.

% Program for the design of Chebyshev Type-1 Bandstop digital lter clc; close all;clear all; format long rp5input(enter the passband ripple...); rs5input(enter the stopband ripple...); wp5input(enter the passband freq...); ws5input(enter the stopband freq...); fs5input(enter the sampling freq...); w152*wp/fs;w252*ws/fs; [n]5cheb1ord(w1,w2,rp,rs); wn5[w1 w2]; [b,a]5cheby1(n,rp,wn,stop); w50:.1/pi:pi; [h,om]5freqz(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.); As an example, enter the passband ripple... 0.25 enter the stopband ripple... 40 enter the passband freq... 2500 enter the stopband freq... 2750 enter the sampling freq... 7000 The amplitude and phase responses of Chebyshev type - 1 bandstop digital lter are shown in Fig. 16.26.

858

Digital Signal Processing

0 0 50

Gain Gain in dB in dB

50 100 100 150 150 200 200 0 0 0.1 0.1 0.2 0.2 0.3 0.3 0.4 0.4 (a) (a) 4 4 3 3 2 0.5 0.5 0.6 0.7 0.8 0.9 0.9 1 1 0.6 0.7 0.8 Normalised frequency Normalised frequency

Phase Phase in radians in radians

1 2 1 0 1 0 2 1 2 3 3 0 0 0.1 0.1 0.2 0.2 0.3 0.3 0.4 0.4 (b) 0.5 0.5 0.6 0.7 0.8 0.9 0.9 1 1

0.6 0.7 0.8 Normalised frequency Normalised frequency

(b) - 1 Bandstop Digital Filter Fig. 16.26 Chebyshev Type (a) Amplitude Response and (b) Phase Response

16.13 CHEBYSHEV TYPE-2 DIGITAL FILTERS

16.13.1 Low-pass Filter Algorithm 1. Get the passband and stopband ripples 2. Get the passband and stopband edge frequencies 3. Get the sampling frequency 4. Calculate the order of the lter using Eq. 8.67 5. Find the lter coefcients 6. Draw the magnitude and phase responses. % Program for the design of Chebyshev Type-2 lowpass digital lter clc; close all;clear all; format long rp5input(enter the passband ripple...); rs5input(enter the stopband ripple...); wp5input(enter the passband freq...); ws5input(enter the stopband freq...); fs5input(enter the sampling freq...); w152*wp/fs;w252*ws/fs; [n,wn]5cheb2ord(w1,w2,rp,rs); [b,a]5cheby2(n,rs,wn);

MATLAB Programs

859

w50:.01:pi; [h,om]5freqz(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.); As an example, enter the passband ripple... 0.35 enter the stopband ripple... 35 enter the passband freq... 1500 enter the stopband freq... 2000 enter the sampling freq... 8000 The amplitude and phase responses of Chebyshev type - 2 low-pass digital lter are shown in Fig. 16.27.
20 20 0 0 20 20 40 40 60 60 80 80 100 100 0 0

Gain Gain in in dBdB

0.1 0.1

0.2 0.2

0.3 0.3

0.4 0.4 (a) (a)

0.5 0.5

0.6 0.7 0.8 0.6 0.7 0.8 Normalised frequency Normalised frequency

0.9 0.9

1 1

4 4 2 2 0 0 2 2 4 4 0 0 0.1 0.1 0.2 0.2 0.3 0.3 0.4 0.4 (b) (b) 0.5 0.5 0.6 0.7 0.8 0.6 0.7 0.8 Normalised frequency Normalised frequency 0.9 0.9 1 1

Fig. 16.27 Chebyshev Type - 2 Low-pass Digital Filter (a) Amplitude Response and (b) Phase Response

16.13.2 High-pass Filter Algorithm 1. Get the passband and stopband ripples 2. Get the passband and stopband edge frequencies

Phase in in radians Phase radians

860

Digital Signal Processing

3. 4. 5. 6.

Get the sampling frequency Calculate the order of the lter using Eq. 8.67 Find the lter coefcients Draw the magnitude and phase responses.

% Program for the design of Chebyshev Type-2 high pass digital lter clc; close all;clear all; format long rp5input(enter the passband ripple...); rs5input(enter the stopband ripple...); wp5input(enter the passband freq...); ws5input(enter the stopband freq...); fs5input(enter the sampling freq...); w152*wp/fs;w252*ws/fs; [n,wn]5cheb2ord(w1,w2,rp,rs); [b,a]5cheby2(n,rs,wn,high); w50:.01/pi:pi; [h,om]5freqz(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.); As an example, enter the passband ripple... 0.25 enter the stopband ripple... 40 enter the passband freq... 1400 enter the stopband freq... 1800 enter the sampling freq... 7000 The amplitude and phase responses of Chebyshev type - 2 high-pass digital lter are shown in Fig. 16.28.
0 20 40

Gain in dB

60 80 100 120 0 0.1 0.2 0.3 0.4 (a) 4 2 0.5 0.6 0.7 0.8 0.9 1 Normalised frequency

Fig. 16.28 (Contd.)

ase in radians

0 2

Gain in

80 100 120 0 0.1 0.2 0.3 0.4 (a) 4 2 0.5 0.6 0.7 0.8 0.9 1 Normalised frequency MATLAB Programs

861

Phase in radians

0 2 4

0.1

0.2

0.3

0.4 (b)

0.5

0.6

0.7

0.8

0.9

Normalised frequency

Fig. 16.28 Chebyshev Type - 2 High-pass Digital Filter (a) Amplitude Response and (b) Phase Response

16.13.3 Bandpass Filter Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequency Get the sampling frequency Calculate the order of the lter using Eq. 8.67 Find the lter coefcients Draw the magnitude and phase responses.

% Program for the design of Chebyshev Type-2 Bandpass digital lter clc; close all;clear all; format long rp5input(enter the passband ripple...); rs5input(enter the stopband ripple...); wp5input(enter the passband freq...); ws5input(enter the stopband freq...); fs5input(enter the sampling freq...); w152*wp/fs;w252*ws/fs; [n]5cheb2ord(w1,w2,rp,rs); wn5[w1 w2]; [b,a]5cheby2(n,rs,wn,bandpass); w50:.01/pi:pi; [h,om]5freqz(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.);

862

Digital Signal Processing

As an example, enter the passband enter the stopband enter the passband enter the stopband enter the sampling

ripple... ripple... freq... freq... freq...

0.4 40 1400 2000 9000

The amplitude and phase responses of Chebyshev type - 2 bandpass digital lter are shown in Fig. 16.29.
100 0

Gain in dB

100 200 300 400 0 0.1 0.2 0.3 0.4 (a) 4 2 0.5 0.6 0.7 0.8 0.9 1 Normalised frequency

Phase in radians

0 2 4

0.1

0.2

0.3

0.4 (b)

0.5

0.6

0.7

0.8

0.9

Normalised frequency

Fig. 16.29 Chebyshev Type - 2 Bandpass Digital Filter (a) Amplitude Response and (b) Phase Response

16.13.4 Bandstop Filter Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter using Eq. 8.67 Find the lter coefcients Draw the magnitude and phase responses.

MATLAB Programs

863

% Program for the design of Chebyshev Type-2 Bandstop digital lter clc; close all;clear all; format long rp5input(enter the passband ripple...); rs5input(enter the stopband ripple...); wp5input(enter the passband freq...); ws5input(enter the stopband freq...); fs5input(enter the sampling freq...); w152*wp/fs;w252*ws/fs; [n]5cheb2ord(w1,w2,rp,rs); wn5[w1 w2]; [b,a]5cheby2(n,rs,wn,stop); w50:.1/pi:pi; [h,om]5freqz(b,a,w); m520*log10(abs(h)); an5angle(h); subplot(2,1,1);plot(om/pi,m); ylabel(Gain in dB --.);xlabel((a) Normalised frequency --.); subplot(2,1,2);plot(om/pi,an); xlabel((b) Normalised frequency --.); ylabel(Phase in radians --.); As an example, enter enter enter enter enter the the the the the passband stopband passband stopband sampling ripple... ripple... freq... freq... freq... 0.3 46 1400 2000 8000

The amplitude and phase responses of Chebyshev type - 2 bandstop digital lter are shown in Fig. 16.30.
20 0

Gain in dB

20 40 60 80 0 0.1 0.2 0.3 0.4 (a) 3 2 1 0.5 0.6 0.7 0.8 0.9 1 Normalised frequency

Fig. 16.30 (Contd.)

Phase in radians

0 -1 -2 -3 -4

Gain i

40 60 80 0 0.1 0.2 0.3 0.4 (a) 0.5 0.6 0.7 0.8 0.9 1

864

Digital Signal Processing

Normalised frequency

3 2 1

Phase in radians

0 -1 -2 -3 -4 0 0.1 0.2 0.3 0.4 (b) 0.5 0.6 0.7 0.8 0.9 1

Normalised frequency

Fig. 16.30 Chebyshev Type - 2 Bandstop Digital Filter (a) Amplitude Response and (b) Phase Response

16.14

 FIR FILTER DESIGN USING WINDOW TECHNIQUES

In the design of FIR lters using any window technique, the order can be calculated using the formula given by N= 20 log( d pd s ) 13 14.6( f s f p ) / Fs

where dp is the passband ripple, ds is the stopband ripple, fp is the passband frequency, fs is the stopband frequency and Fs is the sampling frequency. 16.14.1 Rectangular Window Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter Find the window coefcients using Eq. 7.37 Draw the magnitude and phase responses.

% Program for the design of FIR Low pass, High pass, Band pass and Bandstop lters using rectangular window clc;clear all;close all; rp5input(enter the passband ripple); rs5input(enter the stopband ripple); fp5input(enter the passband freq); fs5input(enter the stopband freq); f5input(enter the sampling freq); wp52*fp/f;ws52*fs/f; num5220*log10(sqrt(rp*rs))213;

MATLAB Programs

865

dem514.6*(fs2fp)/f; n5ceil(num/dem); n15n11; if (rem(n,2)50) n15n; n5n21; end y5boxcar(n1); % LOW-PASS FILTER b5r1(n,wp,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,1);plot(o/pi,m);ylabel(Gain in dB --.); xlabel((a) Normalised frequency --.); % HIGH-PASS FILTER b5r1(n,wp,high,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,2);plot(o/pi,m);ylabel(Gain in dB --.); xlabel((b) Normalised frequency --.); % BAND PASS FILTER wn5[wp ws]; b5r1(n,wn,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,3);plot(o/pi,m);ylabel(Gain in dB -->); xlabel((c) Normalised frequency -->); % BAND STOP FILTER b5r1(n,wn,stop,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,4);plot(o/pi,m);ylabel(Gain in dB -->); xlabel((d) Normalised frequency -->); As an example, enter the passband ripple 0.05 enter the stopband ripple 0.04 enter the passband freq 1500 enter the stopband freq 2000 enter the sampling freq 9000 The gain responses of low-pass, high-pass, bandpass and bandstop lters using rectangular window are shown in Fig. 16.31.

866

Digital Signal Processing

20 0

20 0

Gain in dB

40 60 80 0 0.2 0.4 0.6 0.8 Normalised frequency (a) 20 0 1

Gain in dB

20

20 40 60 80 0 0.2 0.4 0.6 0.8 Normalised frequency (b) 5 0 1

Gain in dB

40 60 80 0 0.2 0.4 0.6 0.8 Normalised frequency (c) 1

Gain in dB

20

5 10 15 20 0 0.2 0.4 0.6 0.8 Normalised frequency (d) 1

Fig. 16.31 Filters Using Rectangular Window (a) Low-pass (b) High-pass (c) Bandpass and (d) Bandstop

16.14.2 Bartlett Window Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter Find the lter coefcients Draw the magnitude and phase responses.

% Program for the design of FIR Low pass, High pass, Band pass and Bandstop lters using Bartlett window clc;clear all;close all; rp5input(enter the passband ripple); rs5input(enter the stopband ripple); fp5input(enter the passband freq); fs5input(enter the stopband freq); f5input(enter the sampling freq);

MATLAB Programs

867

wp52*fp/f;ws52*fs/f; num5220*log10(sqrt(rp*rs))213; dem514.6*(fs2fp)/f; n5ceil(num/dem); n15n11; if (rem(n,2)50) n15n; n5n21; end y5bartlett(n1); % LOW-PASS FILTER b5r1(n,wp,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,1);plot(o/pi,m);ylabel(Gain in dB --.); xlabel((a) Normalised frequency --.); % HIGH-PASS FILTER b5r1(n,wp,high,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,2);plot(o/pi,m);ylabel(Gain in dB --.); xlabel((b) Normalised frequency --.); % BAND PASS FILTER wn5[wp ws]; b5r1(n,wn,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,3);plot(o/pi,m);ylabel(Gain in dB --.); xlabel((c) Normalised frequency --.); % BAND STOP FILTER b5r1(n,wn,stop,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,4);plot(o/pi,m);ylabel(Gain in dB --.); xlabel((d) Normalised frequency --.); As an example, enter the passband ripple 0.04 enter the stopband ripple 0.02 enter the passband freq 1500 enter the stopband freq 2000 enter the sampling freq 8000 The gain responses of low-pass, high-pass, bandpass and bandstop lters using Bartlett window are shown in Fig. 16.32.

868

Digital Signal Processing

0 5 10

5 0 5

Gain in dB

Gain in dB
0.2 0.4 0.6 0.8 Normalised frequency (a) 1

15 20 25 30 35 0

10 15 20 25 30 0 0.2 0.4 0.6 0.8 Normalised frequency (b) 2 0 1

0 10

Gain in dB

20 30 40 0

Gain in dB
0.2 0.4 0.6 0.8 Normalised frequency (c) 1

2 4 6 8 0 0.2 0.4 0.6 0.8 Normalised frequency (d) 1

Fig. 16.32 Filters using Bartlett Window (a) Low-pass (b) High-pass (c) Bandpass and (d) Bandstop

16.14.3 Blackman window Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter Find the window coefcients using Eq. 7.45 Draw the magnitude and phase responses.

% Program for the design of FIR Low pass, High pass, Band pass and Band stop digital lters using Blackman window clc;clear all;close all; rp5input(enter the passband ripple); rs5input(enter the stopband ripple); fp5input(enter the passband freq); fs5input(enter the stopband freq); f5input(enter the sampling freq);

MATLAB Programs

869

wp52*fp/f;ws52*fs/f; num5220*log10(sqrt(rp*rs))213; dem514.6*(fs2fp)/f; n5ceil(num/dem); n15n11; if (rem(n,2)50) n15n; n5n21; end y5blackman(n1); % LOW-PASS FILTER b5r1(n,wp,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,1);plot(o/pi,m);ylabel(Gain in dB --.); xlabel((a) Normalised frequency --.); % HIGH-PASS FILTER b5r1(n,wp,high,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,2);plot(o/pi,m);ylabel(Gain in dB --.); xlabel((b) Normalised frequency --.); % BAND PASS FILTER wn5[wp ws]; b5r1(n,wn,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,3);plot(o/pi,m);ylabel(Gain in dB --.); xlabel((c) Normalised frequency --.); % BAND STOP FILTER b5r1(n,wn,stop,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,4);plot(o/pi,m);;ylabel(Gain in dB --.); xlabel((d) Normalised frequency --.); As an example, enter the passband ripple 0.03 enter the stopband ripple 0.01 enter the passband freq 2000 enter the stopband freq 2500 enter the sampling freq 7000 The gain responses of low-pass, high-pass, bandpass and bandstop lters using Blackman window are shown in Fig. 16.33.

870

Digital Signal Processing

20 0 20

50 0

Gain in dB

40 60 80

Gain in dB
0 0.2 0.4 0.6 0.8 Normalised frequency (a) 1

50 100 150

100 120

0.2 0.4 0.6 0.8 Normalised frequency (b)

0 20

2 0

Gain in dB

60 80

Gain in dB
0.2 0.4 0.6 0.8 Normalised frequency (c) 1

40

2 4 6 8 0 0.2 0.4 0.6 0.8 Normalised frequency (d) 1

100 120 0

Fig. 16.33 Filters using Blackman Window (a) Low-pass (b) High-pass (c) Bandpass and (d) Bandstop

16.14.4 Chebyshev Window Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter Find the lter coefcients Draw the magnitude and phase responses.

% Program for the design of FIR Lowpass, High pass, Band pass and Bandstop lters using Chebyshev window clc;clear all;close all; rp5input(enter the passband ripple); rs5input(enter the stopband ripple); fp5input(enter the passband freq); fs5input(enter the stopband freq); f5input(enter the sampling freq);

MATLAB Programs

871

r5input(enter the ripple value(in dBs)); wp52*fp/f;ws52*fs/f; num5220*log10(sqrt(rp*rs))213; dem514.6*(fs2fp)/f; n5ceil(num/dem); if(rem(n,2)50) n5n11; end y5chebwin(n,r); % LOW-PASS FILTER b5r1(n-1,wp,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,1);plot(o/pi,m);ylabel(Gain in dB --.); xlabel((a) Normalised frequency --.); % HIGH-PASS FILTER b5r1(n21,wp,high,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,2);plot(o/pi,m);ylabel(Gain in dB --.); xlabel((b) Normalised frequency --.); % BAND-PASS FILTER wn5[wp ws]; b5r1(n21,wn,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,3);plot(o/pi,m);ylabel(Gain in dB --.); xlabel((c) Normalised frequency --.); % BAND-STOP FILTER b5r1(n21,wn,stop,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,4);plot(o/pi,m);ylabel(Gain in dB --.); xlabel((d) Normalised frequency --.); As an example, enter the passband ripple 0.03 enter the stopband ripple 0.02 enter the passband freq 1800 enter the stopband freq 2400 enter the sampling freq 10000 enter the ripple value(in dBs)40 The gain responses of low-pass, high-pass, bandpass and bandstop lters using Chebyshev window are shown in Fig. 16.34.

872

Digital Signal Processing

20 0

20 0

Gain in dB

Gain in dB
0 0.2 0.4 0.6 0.8 Normalised frequency (a) 1

20 40 60 80 100

20 40 60 80

100

0.2 0.4 0.6 0.8 Normalised frequency (b)

0 20

2 0 2

Gain in dB

Gain in dB
0.2 0.4 0.6 0.8 Normalised frequency (c) 1

40 60 80

4 6 8 10 12 0 0.2 0.4 0.6 0.8 Normalised frequency (d) 1

100 120 0

Fig. 16.34 Filters using Chebyshev Window (a) Low-pass (b) High-pass (c) Bandpass and (d) Bandstop

16.14.5 Hamming Window Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter Find the window coefcients using Eq. 7.40 Draw the magnitude and phase responses.

% Program for the design of FIR Low pass, High pass, Band pass and Bandstop lters using Hamming window clc;clear all;close all; rp5input(enter the passband ripple); rs5input(enter the stopband ripple); fp5input(enter the passband freq); fs5input(enter the stopband freq); f5input(enter the sampling freq);

MATLAB Programs

873

wp52*fp/f;ws52*fs/f; num5220*log10(sqrt(rp*rs))213; dem514.6*(fs2fp)/f; n5ceil(num/dem); n15n11; if (rem(n,2)50) n15n; n5n21; end y5hamming(n1); % LOW-PASS FILTER b5r1(n,wp,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,1);plot(o/pi,m);ylabel(Gain in dB --.); xlabel((a) Normalised frequency --.); % HIGH-PASS FILTER b5r1(n,wp,high,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,2);plot(o/pi,m);ylabel(Gain in dB --.); xlabel((b) Normalised frequency --.); % BAND PASS FILTER wn5[wp ws]; b5r1(n,wn,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,3);plot(o/pi,m);ylabel(Gain in dB --.); xlabel((c) Normalised frequency --.); % BAND STOP FILTER b5r1(n,wn,stop,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,4);plot(o/pi,m);ylabel(Gain in dB --.); xlabel((d) Normalised frequency --.); As an example, enter the passband ripple 0.02 enter the stopband ripple 0.01 enter the passband freq 1200 enter the stopband freq 1700 enter the sampling freq 9000 The gain responses of low-pass, high-pass, bandpass and bandstop lters using Hamming window are shown in Fig. 16.35.

874

Digital Signal Processing

20 0 20

20 0

Gain in dB

40 60 80 100 120 0 0.2 0.4 0.6 0.8 Normalised frequency (a) 0 20 1

Gain in dB

20 40 60 80

100

0.2 0.4 0.6 0.8 Normalised frequency (b)

2 0

Gain in dB

Gain in dB
0.2 0.4 0.6 0.8 Normalised frequency (c)

40 60 80 100 120 0 1

5 10 15

0.2 0.4 0.6 0.8 Normalised frequency (d)

Fig. 16.35 Filters using Hamming Window (a) Low-pass (b) High-pass (c) Bandpass and (d) Bandstop

16.14.6 Hanning Window Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter Find the window coefcients using Eq. 7.44 Draw the magnitude and phase responses.

% Program for the design of FIR Low pass, High pass, Band pass and Band stop lters using Hanning window clc;clear all;close all; rp5input(enter the passband ripple); rs5input(enter the stopband ripple); fp5input(enter the passband freq); fs5input(enter the stopband freq); f5input(enter the sampling freq);

MATLAB Programs

875

wp52*fp/f;ws52*fs/f; num5220*log10(sqrt(rp*rs))213; dem514.6*(fs2fp)/f; n5ceil(num/dem); n15n11; if (rem(n,2)50) n15n; n5n21; end y5hamming(n1); % LOW-PASS FILTER b5r1(n,wp,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,1);plot(o/pi,m);ylabel(Gain in dB --.); xlabel((a) Normalised frequency --.); % HIGH-PASS FILTER b5r1(n,wp,high,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,2);plot(o/pi,m);ylabel(Gain in dB --.); xlabel((b) Normalised frequency --.); % BAND PASS FILTER wn5[wp ws]; b5r1(n,wn,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,3);plot(o/pi,m);ylabel(Gain in dB --.); xlabel((c) Normalised frequency --.); % BAND STOP FILTER b5r1(n,wn,stop,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,4);plot(o/pi,m);ylabel(Gain in dB --.); xlabel((d) Normalised frequency --.); As an example, enter the passband ripple 0.03 enter the stopband ripple 0.01 enter the passband freq 1400 enter the stopband freq 2000 enter the sampling freq 8000 The gain responses of low-pass, high-pass, bandpass and bandstop lters using Hanning window are shown in Fig. 16.36.

876

Digital Signal Processing

20 0 20

20 0

Gain in dB

40 60 80 100 120 0 0.2 0.4 0.6 0.8 Normalised frequency (a) 0 20 1

Gain in dB

20 40 60 80

100

0.2 0.4 0.6 0.8 Normalised frequency (b)

2 0 2

Gain in dB

Gain in dB
0.2 0.4 0.6 0.8 Normalised frequency (c)

40 60 80 100 120 0 1

4 6 8 10 12 0 0.2 0.4 0.6 0.8 Normalised frequency (d) 1

Fig. 16.36 Filters using Hanning Window (a) Low-pass (b) High-pass (c) Bandpass and (d) Bandstop

16.14.7 Kaiser Window Algorithm 1. 2. 3. 4. 5. 6. Get the passband and stopband ripples Get the passband and stopband edge frequencies Get the sampling frequency Calculate the order of the lter Find the window coefcients using Eqs 7.46 and 7.47 Draw the magnitude and phase responses.

% Program for the design of FIR Low pass, High pass, Band pass and Bandstop lters using Kaiser window clc;clear all;close all; rp5input(enter the passband ripple); rs5input(enter the stopband ripple); fp5input(enter the passband freq); fs5input(enter the stopband freq); f5input(enter the sampling freq); beta5input(enter the beta value);

MATLAB Programs

877

wp52*fp/f;ws52*fs/f; num5220*log10(sqrt(rp*rs))213; dem514.6*(fs2fp)/f; n5ceil(num/dem); n15n11; if (rem(n,2)50) n15n; n5n21; end y5kaiser(n1,beta); % LOW-PASS FILTER b5r1(n,wp,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,1);plot(o/pi,m);ylabel(Gain in dB -->); xlabel((a) Normalised frequency -->); % HIGH-PASS FILTER b5r1(n,wp,high,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,2);plot(o/pi,m);ylabel(Gain in dB -->); xlabel((b) Normalised frequency -->); % BAND PASS FILTER wn5[wp ws]; b5r1(n,wn,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,3);plot(o/pi,m);ylabel(Gain in dB -->); xlabel((c) Normalised frequency -->); % BAND STOP FILTER b5r1(n,wn,stop,y); [h,o]5freqz(b,1,256); m520*log10(abs(h)); subplot(2,2,4);plot(o/pi,m);ylabel(Gain in dB -->); xlabel((d) Normalised frequency -->); As an example, enter the passband ripple 0.02 enter the stopband ripple 0.01 enter the passband freq 1000 enter the stopband freq 1500 enter the sampling freq 10000 enter the beta value 5.8 The gain responses of low-pass, high-pass, bandpass and bandstop lters using Kaiser window are shown in Fig. 16.37.

878

Digital Signal Processing

20 0 20

20 0

Gain in dB

60 80 100 120 0 0.2 0.4 0.6 0.8 Normalised frequency (a) 0 20 1

Gain in dB

40

20 40 60 80 0 0.2 0.4 0.6 0.8 Normalised frequency (b) 5 0 1

Gain in dB

Gain in dB
0.2 0.4 0.6 0.8 Normalised frequency (c) 1

40 60 80 100 120 0

5 10 15

0.2 0.4 0.6 0.8 Normalised frequency (d)

Fig. 16.37 Filters using Kaiser Window (a) Low-pass (b) High-pass (c) Bandpass and (d) Bandstop

16.15 UPSAMPLING A SINUSOIDAL SIGNAL

% Program for upsampling a sinusoidal signal by factor L N5input(Input length of the sinusoidal sequence5); L5input(Up Samping factor5); 5input(Input signal frequency5); % Generate the sinusoidal sequence for the specied length N n50:N21; x5sin(2*pi**n); % Generate the upsampled signal y5zeros (1,L*length(x)); y([1:L:length(y)])5x; %Plot the input sequence subplot (2,1,1); stem (n,x); title(Input Sequence); xlabel(Time n); ylabel(Amplitude);

MATLAB Programs

879

%Plot the output sequence subplot (2,1,2); stem (n,y(1:length(x))); title([output sequence,upsampling factor5,num2str(L)]); xlabel(Time n); ylabel(Amplitude);

16.16

 UPSAMPLING AN EXPONENTIAL SEQUENCE

% Program for upsampling an exponential sequence by a factor M n5input(enter length of input sequence ); l5input(enter up sampling factor ); % Generate the exponential sequence m50:n21; a5input(enter the value of a ); x5a.^m; % Generate the upsampled signal y5zeros(1,l*length(x)); y([1:l:length(y)])5x; gure(1) stem(m,x); xlabel({Time n;(a)}); ylabel(Amplitude); gure(2) stem(m,y(1:length(x))); xlabel({Time n;(b)}); ylabel(Amplitude); As an example, enter length of input sentence 25 enter upsampling factor 3 enter the value of a 0.95 The input and output sequences of upsampling an exponential sequence an are shown in Fig. 16.38.

Fig. 16.38 (Contd.)

880

Digital Signal Processing

Fig. 16.38 (a) Input Exponential Sequence (b) Output Sequence Upsampled by a Factor of 3

16.17

 DOWN SAMPLING A SINUSOIDAL SEQUENCE

% Program for down sampling a sinusoidal sequence by a factor M N5input(Input length of the sinusoidal signal5); M5input(Down samping factor5); 5input(Input signal frequency5); %Generate the sinusoidal sequence n50:N21; m50:N*M21; x5sin(2*pi**m); %Generate the down sampled signal y5x([1:M:length(x)]); %Plot the input sequence subplot (2,1,1); stem(n,x(1:N)); title(Input Sequence); xlabel(Time n); ylabel(Amplitude); %Plot the down sampled signal sequence subplot(2,1,2); stem(n,y); title([Output sequence down sampling factor,num2str(M)]); xlabel(Time n); ylabel(Amplitude);

16.18

 DOWN SAMPLING AN EXPONENTIAL SEQUENCE

% Program for downsampling an exponential sequence by a factor M N5input(enter the length of the output sequence ); M5input(enter the down sampling factor );

MATLAB Programs

881

% Generate the exponential sequence n50:N21; m50:N*M21; a5input(enter the value of a ); x5a.^m; % Generate the downsampled signal y5x([1:M:length(x)]); gure(1) stem(n,x(1:N)); xlabel({Time n;(a)}); ylabel(Amplitude); gure(2) stem(n,y); xlabel({Time n;(b)}); ylabel(Amplitude); As an example, enter the length of the output sentence 25 enter the downsampling factor 3 enter the value of a 0.95 The input and output sequences of downsampling an exponential sequence an are shown in Fig. 16.39.

Fig. 16.39 (a) Input Exponential Sequence (b) Output Sequence Downsampled by a Factor of 3

882

Digital Signal Processing

16.19 DECIMATOR
% Program for downsampling the sum of two sinusoids using MATLABs inbuilt decimation function by a factor M N5input(Length of the input signal5); M5input(Down samping factor5); f15input(Frequency of rst sinusoid5); f25input(Frequency of second sinusoid5); n50:N21; % Generate the input sequence x52*sin(2*pi*f1*n)13*sin(2*pi*f2*n); %Generate the decimated signal % FIR low pass decimation is used y5decimate(x,M,r); %Plot the input sequence subplot (2,1,1); stem (n,x(1:N)); title(Input Sequence); xlabel(Time n); ylabel(Amplitude); %Plot the output sequence subplot (2,1,2); m50:N/M21; stem (m,y(1:N/M)); title([Output sequence down sampling factor,num2str(M)]); xlabel(Time n); ylabel(Amplitude);

16.20 DECIMATOR AND INTERPOLATOR

% Program for downsampling and upsampling the sum of two sinusoids using MATLABs inbuilt decimation and interpolation function by a factor of 20. %Generate the input sequence for Fs5200Hz, f1550Hz and f25100 Hz t50:1/200:10; y53.*cos(2*pi*50.*t/200)11.*cos(2*pi*100.*t/200); gure(1) stem(y); xlabel({Times in Seconds;(a)}); ylabel(Amplitude);

MATLAB Programs

883

%Generate the decimated and interpolated signals gure(2) stem(decimate(y,20)); xlabel({Times in Seconds;(b)}); ylabel(Amplitude); gure(3) stem(interp(decimate(y,20),2)); xlabel({Times in Seconds;(c)}); ylabel(Amplitude);
5 Amplitude

0 5

500

1000 1500 Time in Seconds (a)

2000

2500

5 Amplitude

0 5

20

40

60 Time in Seconds (b)

80

100

120

5 Amplitude

0 5

50

100 150 Time in Seconds (c)

200

250

Fig. 16.40 (a) Input Sequence, (b) Decimated Sequence and (c) Interpolated sequence

16.21

 ESTIMATION OF POWER SPECTRAL DENSITY (PSD)

% Program for estimating PSD of two sinusoids plus noise % % % % Algorithm; 1:Get the frequencies of the two sinusoidal waves 2:Get the sampling frequency 3:Get the length of the sequence to be considered

884

Digital Signal Processing

% 4: Get the two FFT lengths for comparing the corresponding power spectral densities clc; close all; clear all; f15input(Enter the frequency of rst sinusoid); f25input(Enter the frequency of second sinusoid); fs5input(Enter the sampling frequency); N5input(Enter the length of the input sequence); N15input(Enter the input FFT length 1); N25input(Enter the input FFT length 2); %Generation of input sequence t50:1/fs:1; x52*sin(2*pi*f1*1)13*sin(2*pi*f2*t)2randn(size(t)); %Generation of psd for two different FFT lengths Pxx15abs(fft(x,N1)).^2/(N11); Pxx25abs(fft(x,N2)).^2/(N11); %Plot the psd; subplot(2,1,1); plot ((0:(N121))/N1*fs,10*log10(Pxx1)); xlabel(Frequency in Hz); ylabel(Power spectrum in dB); title([PSD with FFT length,num2str(N1)]); subplot (2,1,2); plot ((0:(N221))/N2*fs,10*log10(Pxx2)); xlabel(Frequency in Hz); ylabel(Power spectrum in dB); title([PSD with FFT length,num2str(N2)]);

16.22 PSD ESTIMATOR

% Program for estimating PSD of a two sinusoids plus noise using %(i)non-overlapping sections %(ii)overlapping sections and averaging the periodograms clc; close all; clear all; f15input(Enter the frequency of rst sinusoid); f25input(Enter the frequency of second sinusoid); fs5input(Enter the sampling frequency); N5input(Enter the length of the input sequence); N15input(Enter the input FFT length 1); N25input(Enter the input FFT length 2); %Generation of input sequence t50:1/fs:1; x52*sin(2*pi*f1*1)13*sin(2*pi*f2*t)2randn(size(t));

MATLAB Programs

885

%Generation of psd for two different FFT lengths Pxx15(abs(fft(x(1:256))).^21abs(fft(x(257:512))).^21 abs(fft(x(513:768))).^2/(256*3); %using nonoverlapping sections Pxx25(abs(fft(x(1:256))).^21abs(fft(x(129:384))).^21ab s(fft(x(257:512))).^21abs(fft(x(385:640))).^21abs(fft( x(513:768))).^21abs(fft(x(641:896))).^2/(256*6); %using overlapping sections % Plot the psd; subplot (2,1,1); plot ((0:255)/256*fs,10*log10(Pxx1)); xlabel(Frequency in Hz); ylabel(Power spectrum in dB); title([PSD with FFT length,num2str(N1)]); subplot (2,1,2); plot ((0:255)/256*fs,10*log10(Pxx2)); xlabel(Frequency in Hz); ylabel(Power spectrum in dB); title([PSD with FFT length,num2str(N2)]);

16.23 PERIODOGRAM ESTIMATION

% Periodogram estimate cold be done by applying a nonrectangular data windows to the sections prior to computing the periodogram % This program estimates PSD for the input signal of two sinusoids plus noise using Hanning window f15input(Enter the frequency of rst sinusoid); f25input(Enter the frequency of second sinusoid); fs5input(Enter the sampling frequency); t50:1/fs:1; w5hanning(256); x52*sin(2*pi*f1*t)13*sin(2*pi*f2*t)2randn(size(t)); Pxx5(abs(fft(w.*x(1:256))).^21abs(fft(w.*x(129:384))).^ 21abs(fft(w.*x(257:512))).^21abs(fft(w.*x(385:640))).^2 1abs(fft(w.*x(513:768))).^21abs(fft(w.*x(641:896))).^2/ (norm(w)^2*6); Plot((0:255)/256*fs,10*log10(Pxx));

16.24 WELCH PSD ESTIMATOR

% Program for estimating the PSD of sum of two sinusoids plus noise using Welch method n50.01:0.001:.1; x5sin(.25*pi*n)13*sin(.45*pi*n)1rand(size(n)); pwelch(x)

886

Digital Signal Processing

Power Spectrum Density (dB/rad/sample)

5 0 5 10 15

Welch PSD Estimate

20

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Normalised Frequency (x pi rad/sample)

Fig. 16.41 Welch PSD Estimate

xlabel(Normalised Frequency (x pi rad/sample)); ylabel(Power Spectrum Density (dB/rad/sample))

WELCH PSD ESTIMATOR USING WINDOWS 16.25 

% Program for estimating the PSD of sum of two sinusoids using Welch method with an overlap of 50 percent and with Hanning, Hamming, Bartlett, Blackman and rectangular windows. fs51000; t50:1/fs:3; x5sin(2*pi*200*t)1sin(2*pi*400*t); gure(1) subplot(211) pwelch(x,[],[],[],fs); title(Overlay plot of 50 Welch estimates) subplot(212) pwelch(x,hanning(512),0,512,fs) title(N5512 Overlap550% Hanning) gure(2) subplot(211) pwelch(x,[],[],[],fs); title(Overlay plot of 50 Welch estimates) subplot(212) pwelch(x,hamming(512),0,512,fs) title(N5512 Overlap550% Hamming) gure(3) subplot(211) pwelch(x,[],[],[],fs); title(Overlay plot of 50 Welch estimates) subplot(212) pwelch(x,bartlett(512),0,512,fs) title(N5512 Overlap550% Bartlett) gure(4) subplot(211) pwelch(x,[],[],[],fs);

MATLAB Programs

887

title(Overlay plot of 50 Welch estimates) subplot(212) pwelch(x,blackman(512),0,512,fs) title(N5512 Overlap550% Blackman) gure(5) subplot(211) pwelch(x,[],[],[],fs); title(Overlay plot of 50 Welch estimates) subplot(212) pwelch(x,boxcar(512),0,512,fs) title(N5512 Overlap550% Rectangular)
Power Spectral Density (dB/Hz)

0
20 40 60 80

Overlay plot of 50 Welch estimates

50

100

150

Power Spectral Density (dB/Hz)

0
50

200 250 300 350 Frequency (Hz) N=512 Overlap = 50% Hanning

400

450

500

100

150

50

100

150

200 250 300 Frequency (Hz)

350

400

450

500

Fig. 16.42 (a) Welch Estimate with N5512, 50% Overlap Hanning
Power Spectral Density (dB/Hz)

0
20 40 60 80

Overlay plot of 50 Welch estimates

50

100

150

Power Spectral Density (dB/Hz)

0
20 40 60 80

200 250 300 350 Frequency (Hz) N=512 Overlap = 50% Hamming

400

450

500

50

100

150

200 250 300 Frequency (Hz)

350

400

450

500

Fig. 16.42 (b) Welch Estimate with N5512, 50% Overlap Hamming

888

Digital Signal Processing

Power Spectral Density (dB/Hz)

0
20 40 60 80

Overlay plot of 50 Welch estimates

50

100

150

Power Spectral Density (dB/Hz)

0
20 40 60 80

200 250 300 350 Frequency (Hz) N=512 Overlap = 50% Bartlett

400

450

500

50

100

150

200 250 300 Frequency (Hz)

350

400

450

500

Fig. 16.42 (c) Welch Estimate with N5512, 50% Overlap Bartlett

Power Spectral Density (dB/Hz)

0
20 40 60 80

Overlay plot of 50 Welch estimates

50

100

150

Power Spectral Density (dB/Hz)

0
50 100 150 200 0

200 250 300 350 Frequency (Hz) N=512 Overlap = 50% Blackman

400

450

500

50

100

150

200 250 300 Frequency (Hz)

350

400

450

500

Fig. 16.42 (d) Welch Estimate with N5512, 50% Overlap Blackman

MATLAB Programs Overlay plot of 50 Welch estimates

889

Power Spectral Density (dB/Hz)

0
20 40 60 80

50

100

150

200 250 300 Frequency (Hz)

350

400

450

500

Power Spectral Density (dB/Hz)

0
10 20 30 40 50 60 0 50 100

N = 512 Overlap = 50% Rectangular

150

250 300 200 Frequency (Hz)

350

400

450

500

Fig. 16.42 (e) Welch Estimate with N5512, 50% Overlap Rectangular

16.26 WELCH PSD ESTIMATOR USING WINDOWS

% Program for estimating the PSD of sum of two sinusoids plus noise using Welch method with an overlap of 50 percent and with Hanning, Hamming, Bartlett, Blackman and rectangular windows fs51000; t50:1/fs:3; x52*sin(2*pi*200*t)15*sin(2*pi*400*t); y5x1randn(size(t)); gure(1) subplot(211); pwelch(y,[],[],[],fs); title(Overlay plot of 50 Welch estimates); subplot(212); pwelch(y,hanning(512),0,512,fs); title(N5512 Overlap550% Hanning); gure(2) subplot(211); pwelch(y,[],[],[],fs); title(Overlay plot of 50 Welch estimates); subplot(212); pwelch(y,hamming(512),0,512,fs); title(N5512 Overlap550% Hamming);

890

Digital Signal Processing

gure(3) subplot(211); pwelch(y,[],[],[],fs); title(Overlay plot of 50 Welch estimates); subplot(212); pwelch(y,bartlett(512),0,512,fs); title(N5512 Overlap550% Bartlett); gure(4) subplot(211); pwelch(y,[],[],[],fs); title(Overlay plot of 50 Welch estimates); subplot(212); pwelch(y,blackman(512),0,512,fs); title(N5512 Overlap550% Blackman); gure(5) subplot(211); pwelch(y,[],[],[],fs); title(Overlay plot of 50 Welch estimates); subplot(212); pwelch(y,boxcar(512),0,512,fs); title(N5512 Overlap550% Rectangular);

Power Spectral Density (dB/Hz)

10 0
10 20 30 40 0 50 100

Overlay plot of 50 Welch estimates

150

250 300 200 Frequency (Hz)

350

400

450

500

Power Spectral Density (dB/Hz)

10 0
10 20 30 40 0 50 100

N=512 Overlap = 50% Hanning

150

250 300 200 Frequency (Hz)

350

400

450

500

Fig. 16.43 (a) Welch Estimate with N5512, 50% Overlap Hanning

MATLAB Programs

891

Power Spectral Density (dB/Hz)

10 0
10 20 30 40 0 50 100

Overlay plot of 50 Welch estimates

150

250 300 200 Frequency (Hz)

350

400

450

500

Power Spectral Density (dB/Hz)

10 0
10 20 30 40 0 50 100

N=512 Overlap = 50% Hanning

150

250 300 200 Frequency (Hz)

350

400

450

500

Fig. 16.43 (b) Welch Estimate with N5512, 50% Overlap Hamming

Power Spectral Density (dB/Hz)

10 0
10 20 30 40 0 50 100

Overlay plot of 50 Welch estimates

150

250 300 200 Frequency (Hz)

350

400

450

500

Power Spectral Density (dB/Hz)

10 0
10 20 30 40 0 50 100 150

N=512 Overlap = 50% Bartlett

250 300 200 Frequency (Hz)

350

400

450

500

Fig. 16.43 (c) Welch Estimate with N5512, 50% Overlap Bartlett

892

Digital Signal Processing

Power Spectral Density (dB/Hz) Power Spectral Density (dB/Hz)

10 0
10 20 30 40 0 50 100

Overlay plot of 50 Welch estimates

150

10 0
10 20 30 40 0 50 100

200 250 300 350 Frequency (Hz) N=512 Overlap = 50% Blackman

400

450

500

150

200 250 300 Frequency (Hz)

350

400

450

500

Fig. 16.43 (d) Welch Estimate with N5512, 50% Overlap Blackman

Power Spectral Density (dB/Hz)

10 0
10 20 30 40 0 50 100

Overlay plot of 50 Welch estimates

150

250 300 200 Frequency (Hz)

350

400

450

500

Power Spectral Density (dB/Hz)

10 0
10 20 30 40 0 50 100

N=512 Overlap = 50% Hanning

150

250 300 200 Frequency (Hz)

350

400

450

500

Fig. 16.43 (e) Welch Estimate with N5512, 50% Overlap Rectangular

MATLAB Programs

893

16.27 STATE-SPACE REPRESENTATION

% Program for computing the state-space matrices from the given transfer function function [A,B,C,D]5tf2ss(b,a); a5input (enter the denominator polynomials5); b5input (enter the numerator polynomials5); p5length(a)21;q5length(b)21;N5max(p,q); if(Np),a5[a,zeros(1,N2p)];end if(Nq),b5[b,zeros(1,N2q)];end A5[2a(2:N11);[eye(N21),zeros(N21,1)]]; B5[1;zeros(N21,1)]; C5b(2:N11)2b(1)*(2:N11); D5b(1);

16.28 PARTIAL FRACTION DECOMPOSITION

% Program for partial fraction decomposition of a rational transfer function function[c,A,alpha]5tf2pf(b,a); a5input (enter the denominator polynomials5); b5input (enter the numerator polynomials5); p5length(a)21; q5length(b)21; a5(1/a(1))*reshape(a,1,p11); b5(1/a(1))*reshape(b,1,q11); if(q5p),%case of nonempty c(z) temp5toeplitz([a,zeros(1,q2p)],[a(1),zeros(1,q2p)]); temp5[temp,[eye(p);zeros(q2p11,p)]); temp5temp/b; c5temp(1:;q2p11); d5temp(q2p12:q11); else c5[]; d5[b,zeros(1,p2q21)]; end alpha5cplxpair (roots(a));; A5zeros(1,p); for k51 :p temp5prod(alpha(k)2alpha(nd(1:p5k))); if(temp550),error(repeated roots in TF2PF); else,A(k)5polyval(d,alpha(k))/temp; end end

894

Digital Signal Processing

16.29 INVERSE z-TRANSFORM

% Program for computing inverse z-transform of a rational transfer function function x5invz(b,a,N); b5input (enter the numerator polynomials5); a5input (enter the denominator polynomials5); N5input (enter the number of points to be computed5); [c,A,alpha]5tf2pf(b,a); x5zeros (1,N); x(1:length(c))5c; for k51:length(A), x5x1A(k)*(alpha(k)).^(0:N21); end x5real(x);

16.30 GROUP DELAY

% Program for computing group delay of a rational transfer function on a given frequency interval function D5grpdly(b,a,K,theta); b5input (enter the numerator polynomials5); a5input (enter the denominator polynomials5); K5input (enter the number of frequency response points5); theta5input (enter the theta value5); a5reshape(a,1,length(a)); b5reshape(b,1,length(b)); if (length(a)551)%case of FIR bd52j*(0:length(b)21).*b; if(nargin553), B5frqresp(b,1,K); Bd5frqresp(bd,1,K); else, B5frqresp(b,1,K,theta); Bd5frqresp(bd,1,K,theta); end D5(real(Bd).*imag(B)2real(B).*imag(Bd))./abs(B).^2; else %case of IIR if(nargin553), D5grpdly (b,1,K)2grpdly(a,1,K); else, D5grpdly(b,1,K,theta)2grpdly(a,1,K,theta); end end

MATLAB Programs

895

16.31

 IIR FILTER DESIGN-IMPULSE INVARIANT METHOD

% Program for transforming an analog lter into a digital lter using impulse invariant technique function [bout,aout]5impinv(bin,ain,T); bin5input(enter the numerator polynomials5); ain5input(enter the denominator polynomials5); T5input(enter the sampling interval5); if(length(bin)5length(ain)), error(Anlog lter in IMPINV is not strictly proper); end [r,p,k]5residue(bin,ain); [bout,aout]5pf2tf([],T*r,exp(T*p));

16.32

 IIR FILTER DESIGN-BILINEAR TRANSFORMATION

% Program for transforming an analog lter into a digial lter using bilinear transformation function [b,a,vout,uout,Cout]5bilin(vin,uin,Cin,T); pin5input(enter the poles5); zin5input(enter the zero5); T5input(enter the sampling interval5); Cin5input(enter the gain of the analog lter5); p5length(pin); q5length(zin); Cout5Cin*(0.5*T)^(p2q)*prod(120.5*T*zin)/ prod(120.5*T*pin); zout5[(110.5*T*zin)./(120.5*T*pin),2ones(1,p2q)]; pout5(110.5*T*pin)./(120.5*T*pin); a51; b51; for k51 :length(pout),a5conv(a,[1,2pout(k)]); end for k51 :length(zout),b5conv(b,[1,2zout(k)]); end a5real(a); b5real(Cout*b); Cout5real(Cout);

16.33

 DIRECT REALISATION OF IIR DIGITAL FILTERS

% Program for computing direct realisation values of IIR digital lter function y5direct(typ,b,a,x); x5input(enter the input sequence5);

896

Digital Signal Processing

b5input(enter the numerator polynomials5); a5input(enter the denominator polynomials5); typ5input(type of realisation5); p5length(a)21; q5length(b)21; pq5max(p,q); a5a(2:p11);u5zeros(1,pq);%u is the internal state; if(typ551) for i51:length(x), unew5x(i)2sum(u(1:p).*a); u5[unew,u]; y(i)5sum(u(1:q11).*b); u5u(1:pq); end elseif(typ552) for i51:length(x) y(i)5b(1)*x(i)1u(1); u5u[(2:pq),0]; u(1:q)5u(1:q)1b(2:q11)*x(i); u(1:p)5u(1:p)2a*y(i); end end

16.34

 PARALLEL REALISATION OF IIR DIGITAL FILTERS

% Program for computing parallel realisation values of IIR digital lter function y5parallel(c,nsec,dsec,x); x5input(enter the input sequence5); b5input(enter the numerator polynomials5); a5input(enter the denominator polynomials5); c5input(enter the gain of the lter5); [n,m]5size(a);a5a(:,2:3); u5zeros(n,2); for i51:length(x), y(i)5c*x(i); for k51:n, unew5x(i)2sum(u(k,:).*a(k,:));u(k,:)5[unew,u(k,1)]; y(i)5y(i)1sum(u(k,:).*b(k,:)); end end

MATLAB Programs

897

16.35

 CASCADE REALISATION OF IIR DIGITAL FILTERS

% Program for computing cascade realisation values of digital IIR lter function y5cascade(c,nsec,dsec,x); x5input(enter the input sequence5); b5input(enter the numerator polynomials5); a5input(enter the denomiator polynomials5); c5input(enter the gain of the lter5); [n,m]5size(b); a5a(:,2:3);b5b(:,2,:3); u5zeros(n,2); for i51 :length(x), for k51 :n, unew5x(i)2sum(u(k,:).*a(k,:)); x(i)52unew1sum(u(k,:).*b(k,:)) u(k,:)5[unew,u(k,1)]; end y(i)5c*x(i); end

16.36

 DECIMATION BY POLYPHASE DECOMPOSITION

% Program for computing convolution and m-fold decimation by polyphase decomposition function y5ppdec(x,h,M); x5input(enter the input sequence5); h5input(enter the FIR lter coefcients5); M5input(enter the decimation factor5); 1h5length(h); 1p5oor((1h21)/M)11; p5reshape([reshape(h,1,1h),zeros(1,1p*M21h)],M,1p); lx5length(x); ly5oor ((1x11h22)/M)11; 1u5foor((1x1M22)/M)11; %length of decimated sequences u5[zeros(1,M21),reshape(x,1,1x),zeros(1,M*lu2lx2M11)]; y5zeros(1,1u11p21); for m51:M,y5y1conv(u(m,: ),p(m,: )); end y5y(1:1y);

16.37 MULTIBAND FIR FILTER DESIGN

% Program for the design of multiband FIR lters function h5rdes(N,spec,win); N5input(enter the length of the lter5);

898

Digital Signal Processing

spec5input(enter the low,high cutoff frequencies and gain5); win5input(enter the window length5); ag5rem(N,2); [K,m]5size(spec); n5(0:N)2N/2; if (ag),n(N/211)51; end,h5zeros(1,N11); for k51:K  temp5(spec (k,3)/pi)*(sin(spec(k,2)*n)2sin(spec(k,1) *n))./n;  if (ag);temp(N/211)5spec(k,3)*(spec(k,2)2 spec(k,1))/pi; end h5h1temp; end if (nargin553), h5h.*reshape(win,1,N11); end

16.38 ANALYSIS FILTER BANK

% Program for maximally decimated uniform DFT analysis lter bank function u5dftanal(x,g,M); g5input(enter the lter coefcient5); x5input(enter the input sequence5); M5input(enter the decimation factor5); 1g5length(g); 1p5oor((1g21)/M)11; p5reshape([reshape(g,1,1g),zeros(1,1p*M21g)],M,1p); lx5length(x); lu5oor ((1x1M22)/M)11; x5[zeros(1,M21),reshape(x,1,1x),zeros(1,M*lu2lx2M11)]; x5ipud(reshape(x,M,1u)); %the decimated sequences u5[]; for m51:M,u5[u; cov(x(m,:),p(m,:))]; end u5ifft(u);

16.39 SYNTHESIS FILTER BANK

% Program for maximally decimated uniform DFT synthesis lter bank function y5udftsynth(v,h,M); 1h5length(h); 1q5oor((1h21)/M)11; q5ipud(reshape([reshape(h,1,1h),zeros(1,1q*M21h)],M,1q));

MATLAB Programs

899

v5fft(v); y5[ ]; for m51:M,y5[conv(v(m,:),q(m,:));y]; end y5y(:).;

16.40 LEVINSON-DURBIN ALGORITHM

% Program for the solution of normal equations using LevinsonDurbin algorithm function [a,rho,s]5levdur(kappa); % Input; % kappa: covariance sequence values from 0 to p % Output parameters: % a: AR polynomial,with leading entry 1 % rho set of p reection coefcients % s: innovation variance kappa5input(enter the covariance sequence5); p5length(kappa)21; kappa5reshape(kappa,p11,1); a51; s5kappa(1); rho5[]; for i51:p, rhoi5(a*kappa(i11:21:2))/s; rho5[rho,rhoi]; s5s*(12rhoi^2); a5[a,0]; a5a2rhoi*iplr(a); end

16.41 WIENER EQUATIONS SOLUTION

% Program function b5wiener(kappax,kappayx); kappax5input(enter the covariance sequence5); kappyx5input(enter the joint covariance sequence5); q5length(kappax)21; kappax5reshape(kappax,q11,1); kappayx5reshape(kappayx,q11,1); b5(toeplitz(kappax)/(kappayx);

16.42 SHORT-TIME SPECTRAL ANALYSIS

% Program function X5stsa(x,N,K,L,w,opt,M,theta0,dtheta); x5input(enter the input signal5); L5input(enter the number consecutive DFTs to average5); N5input(enter the segment length5); K5input(enter the number of overlapping points5); w5input(enter

900

Digital Signal Processing

the window coefcients5); opt5input(opt5); M5input(enter the length of DFT5); theta05input(theta05); dtheta5input(dtheta5); 1x5length(x); nsec5ceil((1x2N)/(N2K)11; x5[reshape(x,1,1x),zeros(1,N1(nsec21)*(N2K))2lx)]; nout5N; if (nargin 5),nout5M; else,opt5n; end X5zeros(nsec,nout); for n51: nsec, temp5w.*x((n21)*(N2K)11:(n21)*(N2K)1N); if (opt(1) 55 z),temp5[temp,zeros(1,M2N)]; end if (opt(1)55c),temp5chirpf (temp,theta0,dtheta,M); else,temp5fftshift(fft(temp)); end X(n,: )5abs(temp).^2; end if(L1); nsecL5oor(nsec/L); for n51:nsecL,X(n,:)5mean (X((n21)*L11:n*L,:)); end if (nsec55nsecL*L11), X(nsecL11,:)5X(nsecL*L11,:); X5X(1:nsecL11),: ); elseif(nsec nsecL*L), X(nsecL11,:)5mean(x(nsecL*L11:nsec,:)); X5X(1:nsecL11,:); else,X5X(1:nsecL,:); end end LKh

16.43
Desired sequence

 CANCELLATION OF ECHO PRODUCED ON THE TELEPHONE-BASE BAND CHANNEL

Base band transmit filter Echo Canceler Echo path

Estimated sequence

Fig. 16.44 Baseband Channel Echo Canceler

% Simulation program for baseband echo cancellation shown in Fig. 16.44 using LMS algorithm clc; close all; clear all; format short T5input(Enter the symbol interval); br5input(Enter the bit rate value); rf5input(Enter the roll off factor); n5[210 10]; y55000*rcosr(rf,n,br,T); %Transmit lter pulse shape is assumed as raised cosine

MATLAB Programs

901

ds5[5 2 5 2 5 2 5 2 5 5 5 5 2 2 2 5 5 5 5]; % data sequence m5length(ds); nl5length(y); i51; z5conv(ds(i),y); while(i) z15[z, zeros(1,1.75*br)]; z5conv(ds(i11),y); z25[zeros(1,i*1.75*br),z]; z5z11z2; i5i11; end %plot(z); %near end signal h5randn(1,length(ds)); %echo path impulse response rs15lter(h,1,z); for i51; length(ds); rs(i)5rs 1(i)/15; end for i51: round(x3/3), rs(i)5randn(1); % rs2echo produced in the hybrid end fs5[5 5 2 2 2 2 2 5 2 2 2 5 5 5 2 5 2 5 2]; % Desired data signal m5length(ds); nl5length(y); i51; z5conv(fs(i),y); while(i) z15[z,zeros(1,1.75*br)]; z5conv(fs(i11),y); z25[zeros(1,i*1.75*br),z]; z5z11z2; i5i11; end fs15rs1fs; % echo added with desired signal ar5xcorr(ds,ds); crd5xcorr(rs,ds); ll5length(ar); j51; for i5round(11/2): 11, ar1(j)5ar(i); j5j11; end r5toeplitz(ar1); l25length(crd); j51; for i5round(l2/2):12, crdl(j)5crd(i); j5j11; end p5crd1;

902

Digital Signal Processing

lam5max(eig(r)); la5min(eig(r)); l5lam/la; w5inv(r)*p; % Initial value of lter coefcients e5rs2lter(w,l,ds); s51; mu51.5/lam; ni51; while (s 1 e210) w15w22*mu*(e.*ds) ; % LMS algorithm adaptation rs y45lter(w1,1,ds); % Estimated echo signal using LMS algorithm e5y42rs; s50; e15xcorr(e); for i51:length(e1), s5s1e1(i); end s5s/length(e1); if (y455rs) break end ni5ni11; w5w1; end gure(1); subplot(2,2,1); plot(z); title(near end signal); subplot(2,2,2); plot(rs); title(echo produced in the hybrid); subplot(2,2,3); plot(fs); title(desired signal); subplot(2,2,4); plot(fs1); title(echo added with desired signal); gure(2); subplot(2,1,1); plot(y4); title(estimated echo signal using LMS algorithm); subplot(2,1,2); plot(fs12y4); title(echo cancelled signal);

16.44

 CANCELLATION OF ECHO PRODUCED ON THE TELEPHONEPASS BAND CHANNEL

Pass band transmit filter Echo Canceler Echo path

Desired sequence

Estimated sequence

Passband receive filter

Fig. 16.45 Pass Band Channel Echo Canceler

MATLAB Programs

903

% Simulation program for passband echo cancellation shown in Fig. 16.45 using LMS algorithm clc; close all; clear all; format long fd58000; fs516000; fc58000; f54000; t50:.01:1; %d5sin(2*pi*f*t/fd); % Near end signal ns5[5 2 5 2 5 5 2 2 2 5 5 2 2 2 2 2 2 5 5 2 5 2 5 5 5 5 5 5 5 5 5 5 5 5 5 5]; % Near end input signal is digitally modulated and plotted y5dmod(ns,fc,fd,fs,psk); subplot(2,2,1); plot(y); title(input signal); xlabel(Time ); ylabel(Amplitude ); % Echo is generated due to mismatch of hybrid impedances h55*randn(1,length(ns)); rsl5lter(h,1,y); % for i51; length(ns); % rsl(i)5rs6(i); % end for i51; length(ns); rs(i)5rs1(i); end subplot(2,2,2); plot(rs); title(noise signal); xlabel(Time ); ylabel(Amplitude ); % Far end signal fs15[5 5 2 5 2 5 2 5 2 5 2 5 2 5 5 5 5 5 5 5 5 5 2 2 2 2 2 2 2 2 2 2 2 2 2 5]; % rs5sign(rs2); % Far end signal is digitally modulated and plotted z15dmod(fs1,fc,fd,fs,psk); for i51:length(ns), z(i)5z1(i); end subplot(2,2,3); plot(z); title(far-end signal); xlabel(Time ); ylabel(Amplitude ); % Echo and the far end modulated signal is added in the hybrid q15z11rs1; for i51; length(ns); q(i)5q1(i);; end subplot(2,2,4); plot(q); title(received signal); xlabel(Time ); ylabel(Amplitude ); q25xcorr(q); % Auto correlation is taken for the near end signal ar5xcorr(ns); % cross correction is taken for the near end and far end signal crd5xcorr(rs,ns);

904

Digital Signal Processing

l15length(ar); j51; for i5round(ll/2): l1, ar1(j)5ar(i) j5j11; end % Toeplitz matrix is taken for the auto correlated signal r5toeplitz(ar1); l25length(crd); j51; for i5round(l2/2):l2, crd1(j)5crd(i); j5j11; end p5crd1; % Maximum and minimum eigen values are calculated from the toeplitz matrix lam5max(eig(r)); la5min(eig(r)); l5lam/la; % initial lter taps are found using the below relation w5inv(r)*p; % The step size factor is calculated m5length(ns)22.5; a5(m2.95367)/.274274; mu5a/lam; % The initial error is calculated s51; e5rs2lter(w,1,ns); ni51; gure(2); subplot(2,2,1); % Filter taps are iterated until the mean squared error becomes E225 while (s25 ! s0) w15w22*mu*(e.*ns); if (ni5100) break; end rs y45lter(w1,1,ns) e5y42rs; s50; el5e.*e; for i51: length(e1), s5s1e1(i); end s5s/length(e1); ni5ni11; w5w1; plot (ni,e); hold on; title( MSE vs no. of iterations); end end subplot(2,2,2); plot(y4); title(estimated noise signal); xlabel(Time ); ylabel(Amplitude ); subplot(2,2,3); plot(q-y4); title(noise cancelled signal); xlabel(Time ); ylabel(Amplitude );

MATLAB Programs

905

Review Questions
16.1 (a) Generate unit impulse function (b) Generate signal x(n)5u(n) 2 u(n 2 N) (c) Plot the sequence, x(n)5A cos ((2p f n)/fs), where n50 to 100 fs5100 Hz, f51 Hz, A50.5 (d) Generate x (n)5exp (25n), where n50 to 10. 16.2 Consider a system with impulse response (1 / 2) n , n = 0 to 4 h( n) = elsewhere. 0, 16.3 Consider the gure.
y (n) ()

( )

h 1 ( n)

h 2 ( n)

h 3 ( n)

h 4 ( n)

Fig. Q16.3

(a) Express the overall impulse response in terms of h1(n), h2(n), h3(n), h4(n) (b) Determine h(n) when h1(n)5{1/2, 1/4, 1/2} h2(n)5h3(n)5(n11) u(n) h4(n)5d(n 2 2) (c) Determine the response of the system in part (b) if x(n)5 d(n12) 13d(n 2 1) 2 4d(n 2 3).

16.4 Compute the overall impulse response of the system

( )
H1 H2
3 (n)

H4

y (n)

H3

y 3(n)

Fig. Q16.4

for 0# n# 99. The system H1, H2, H3, H4, are specied by H1 : h1[n]5{1, 1/2, 1/4, 1/8, 1/16, 1/32} H2 : h2[n]5{1, 1, 1, 1, 1} H3 : y3[n]5(1/4) x(n)1(1/2) x(n 2 1)1(1/4) x(n 2 2) H4 : y[n]50.9 y(n 2 1) 2 0.81 y(n 2 2)1V(n)1V(n 2 1) Plot h(V) for 0#n#99.

16.5 Consider the system with h(n)5an u(n), 21,a,1. Determine the response. x(n)5u(n15) 2 u(n 2 10)

906

Digital Signal Processing

h (n) ()

y (n)

z 2

h(n)

Fig. Q16.5

16.6 (i) Determine the range of values of the parameter a for which the linear time invariant system with impulse response a n , n 0, n even h( n) =  0, otherwise is stable. (ii) Determine the response of the system with impulse response h(n)5anu(n) to the input signal x(n)5u(n) 2 u(n 2 10) 16.7 Consider the system described by the difference equation y(n)5a (y11)1 bx (n). Determine b in terms of a so that S h(n)51. (a) Compute the zero-state step response s(n) of the system and choose b so that s(`)51. (b) Compare the values of b obtained in parts (a) and (b). What did you observe? 16.8 Compute and sketch the convolution y(n) and correlation rxh(n) sequences for the following pair of signals and comment on the results obtained. 1, 2, 4 1, 1, 1, 1, 1 x1 ( n) = h1 ( n) = 0, 1, 2, 3, 4 1 / 2, 1, 2, 1, 1 / 2 x2 ( n) = h2 ( n) = 1, 2, 3, 4 4, 3, 2, 1 x3 ( n) = h3 ( n) = 1 , 2 , 3 , 4 1 , 2 , 3 , 4 x4 ( n) = h4 ( n) = 16.9 Consider the recursive discrete-time system described by the difference equation. y(n)5a1y(n 2 1) 2 a2(n 2 2)1b0 x(n) where a1520.8, a250.64, b050.866 (a) Write a program to compute and plot the impulse response h(n) of the system for 0#n#49. (b) Write a program to compute and plot the zero-state step response s(n) of the system for 0#n#100. (c) Dene an FIR system with impulse response hFIR(n) given by h( n), 0 n 19 hFIR ( n) = elsewhere 0,  where h(n) is the impulse response computed in part (a). Write a program to compute and plot its step response. (d) Compare the results obtained in part (b) and part (c) and explain their similarities and differences.

MATLAB Programs

907

16.10 D  etermine and plot the real and imaginary parts and the magnitude and phase spectra of the following DTFT for various values of r and u G(z)51/(1 2 2r(cos u) z211r2z22) for 0,r,1. 16.11 Using MATLAB program compute the circular convolution of two length-N sequences via the DFT based approach. Using this problem determine the circular convolution of the following pairs of sequences: (a) x(n)5{1, 2 3, 4, 2, 0, 2 2} (b) x(n)5{31j2, 221j, j3, 11j4, 231j3},
n

h(n)5{3, 0, 1, 2 1, 2, 1} h(n)5{ 1 2 j3, 41j2, 2 2 j2, 231j5, 21j }

(c) x(n)5cos (pn/2) h(n)53 0#n#5 16.12 Determine the factored form of the following z-transforms  (a) H1(z)5(2z4116z3144z2156z132)/ (3z313z3 2 15z2118z 2 12) (b) H2(z)5(4z4 2 8.68z3 2 17.98z2126.74z 2 8.04)/

 (z4 2 2z3110z216z165)  and show their pole-zero plots. Determine all regions of convergence of each of the above z-transforms, and describe the type of their inverse z-transform (left-sided, right-sided, two-sided sequences) associated with each of the ROCs. 16.13  Determine the z-transform as a ratio of two polynomials in z21 from each of the partial-fraction expansions listed below: (a) H1 ( z ) = 3 + (b) H 2 ( z ) = 3 + (c) H 3 ( z ) = 12 16 , ( 2 z 1 ) ( 4 z 1 ) z > 0.5

3 ( 4 z 1 ) , (1 + 0.5 z 1 ) (1 + 0.25 z 2 )

z > 0.5 z > 0.4

20 10 4 + , 1 2 1 (5 + 2 z ) (5 + 2 z ) (1 + 0.9 z 2 ) 10 z 1 + , ( 5 + 2 z 1 ) ( 6 + 5 z 1 + z 2 )

(d) H 4 ( z ) = 8 +

z > 0.4

16.14 Determine the inverse z-transform of each z-transform given in Q16.13. 16.15 Consider the system (1 2 z 1 + 2 z 2 z 3 ) H ( z) = , ROC 0.5 < z < 1 (1 z 1 )(1 0.5 z 1 )(1 0.2 z 1 ) (a) Sketch the pole-zero pattern. Is the system stable? (b) Determine the impulse response of the system. 16.16  Determine the impulse response and the step response of the following causal systems. Plot the pole-zero patterns and determine which of the systems are stable. 3 1 (a) y( n) = y( n 1) y( n 2) + x( n) 4 8

908

Digital Signal Processing

(b) y(n)5y (n 2 1) 2 0.5y(n 2 2)1x(n)1x (n 2 1) (c) H ( z ) =

z 1 (1 + z 1 ) (1 z )3 (d) y(n)50.6y(n 2 1) 2 0.08y(n 2 2)1x(n) (e) y(n)50.7y(n 2 1) 2 0.1y(n 2 2)12x(n) 2 x(n 2 2) Ans: (a), (b), (d) and (e) are stable, (c) is unstable 16.17 The frequency analysis of an amplitude-modulated discrete-time signal x(n)5sin 2p f1n1sin 2p f2 n 1 5 where f1 = and f 2 = modulates the amplitude-modulated siganl is 128 128 xc(n) 5 sin 2p fc n where fc 550/128. The resulting amplitude-modulated signal is xam(n)5x(n) sin 2p fc n (a) Sketch the signals x(n), xc(n) and xam(n), 0#n#255 (b) Compute and sketch the 128-point DFT of the signal xam(n), 0#n#127 (c) Compute and sketch the 128-point DFT of the signal xam(n), 0#n#99 (d) Compute and sketch the 256-point DFT of the signal xam(n), 0#n#179 (e) Explain the results obtained in parts (b) through (d) by deriving the spectrum of the amplitude modulated signal and comparing it with the experimental results. 16.18 A continuous time signal xa(t) consists of a linear combination of sinusoidal signals of frequencies 300Hz, 400Hz, 1.3kHz, 3.6KHz and 4.3KHz. The xa(t) is sampled at 4kHz rate and the sampled sequence is passed through an ideal low-pass lter with cut off frequency of 1kHz, generating a continuous time signal ya(t). What are the frequency components present in the reconstructed signal ya(t)? 16.19 Design an FIR linear phase, digital lter approximating the ideal frequency response 1, for / 6 H d ( ) = for / 6 < 0, (a) Determine the coefcient of a 25 tap lter based on the window method with a rectangular window. (b) Determine and plot the magnitude and phase response of the lter. (c) Repeat parts (a) and (b) using the Hamming window (d) Repeat parts (a) and (b) using the Bartlett window. 16.20 Design an FIR Linear Phase, bandstop lter having the ideal frequency response 1, for / 6 H d () = 0, for / 6 < / 3 , for /3 1 (a) Determine the coefcient of a 25 tap lter based on the window method with a rectangular window. (b) Determine and plot the magnitude and phase response of the lter.

MATLAB Programs

909

(c) Repeat parts (a) and (b) using the Hamming window (d) Repeat parts (a) and (b) using the Bartlett window. 16.21 A digital low-pass lter is required to meet the following speccations Passband ripple1 dB Passband edge 4 kHz Stopband attenuation40 dB Stopband edge 6 kHz Sample rate 24 kHz  The lter is to be designed by performing a bilinear transformation on an analog system function. Determine what order Butterworth, Chebyshev and elliptic analog design must be used to meet the specications in the digital implementation. 16.22 An IIR digital low-pass lter is required to meet the following speccations Passband ripple0.5 dB Passband edge 1.2 kHz Stopband attenuation40 dB Stopband edge 2 kHz Sample rate 8 kHz Use the design formulas to determine the lter order for (a) Digital Butterworth lter (b) Digital Chebyshev lter (c) Digital elliptic lter 16.23 An analog signal of the form xa(t)5a(t) cos(2000 t) is bandlimited to the range 900F1100Hz. It is used as an input to the system shown in Fig. Q16.23.
a (t )

A/D R = 2500

( )
cos (0.8 n)

H ( )

(n)

D /A

a(n)

Fig. Q16.23

(a) Determine and sketch the spectra for the signal x(n) and w(n). (b) Use Hamming window of length M531 to design a low-pass linear phase FIR lter H() that passes {a(n)}. (c) Determine the sampling rate of A/D converter that would allow us to eliminate the frequency conversion in the above gure. 16.24 Consider the signal x(n)5an u(m), |a|,1 (a) Determine the spectrum X() (b) The signal x(n) is applied to a device (decimator) which reduces the rate by a factor of two. Determine the output spectrum. (c) Show that the spectrum is simply the Fourier transform of x(2n). 16.25 Design a digital type-I Chebyshev low-pass lter operating at a sampling rate of 44.1kHz with a passband frequency at 2kHz, a pass band ripple of 0.4dB, and a minimum stopband attenuation of 50dB at 12kHz using the impulse invariance method and the bilinear transformation method. Determine the order of analog lter prototype and design the analog prototype lter. Plot the gain and phase responses of the both designs using

910

Digital Signal Processing

MATLAB. Compare the performances of the two lters. Show all steps used in the design. Hint 1. The order of lter N= cosh1 ( ( A2 1) / cosh1 ( s / p) 2. Use the function cheblap. 16.26 Design a linear phase FIR high-pass lter with following specications:  Stopband edge at 0.5p, passband edge at 0.7p, maximum passband attenuation of 0.15dB and a minimum stopband attenuation of 40dB. Use each of the following windows for the design. Hamming, Hanning, Blackman and Kaiser. Show the impulse response coefcients and plot the gain response of the designed lters for each case. 16.27 Design using the windowed Fourier series approach a linear phase FIR lowpass lter with the following specications: pass band edge at 1 rad/s, stop band edge at 2rad/s, maximum passband attenuation of 0.2dB, minimum stopband attenuation of 50dB and a sampling frequency of 10rad/s. Use each of the following windows for the design: Hamming, Hanning, Blackman, Kaiser and Chebyshev. Show the impulse response coefcients and plot the gain response of designed lters for each case. 16.28 Design a two-channel crossover FIR low-pass and high-pass lter pair for digital audio applications. The low-pass and high-pass lters are of length 31 and have a crossover frequency of 2kHz operating at a sampling rate of 44.1KHz. Use the function r1 with a Hamming window to design the lowpass lter while the high-pass lter is derived from the low-pass lter using the delay complementary property. Plot the gain responses of both lters on the same gure. What is the minimum number of delays and multipliers needed to implement the crossover network? 16.29 Design a digital network butterworth low-pass lter operating at sampling rate of 44.1kHz with a 0.5dB cutoff frequency at 2kHz and a minimum stopband attenuation of 45dB at 10kHz using the impulse invariance method and the bilinear transformation method. Assume the sampling interval for the impulse invariance design to be equal to 1. Determine the order of the analog lter prototype and then design the analog prototype lter. Plot the gain and phase responses of both designs. Compare the performances of the lters. Show all steps used in the design. Does the sampling interval have any effect on the design of the digital lter design based on the impulse invariance method? Hint The order of lter is log10 (1 / k1 ) N= log10 (1 / k ) and use the function buttap.