VTU EDUSAT PROGRAM

Lecture on DSP by
Dr K A Radhakrishna Rao
Dept of E & C Engg
PES College of Engg.
Mandya

Goertzels Algorithm
Introduction:
Frequency Analysis
Standard frequency analysis requires transforming time-domain signal to frequency
domain and studying Spectrum of the signal. This is done through DFT computation.
N-point DFT computation results in N frequency components
DFT computation through FFT requires N/2 log2N complex multiplications and
N log2N additions.
In certain applications not all N frequency components need to be computed (an
application will be discussed)
If the desired number of values of the DFT is less than 2 log2N than direct
computation of the desired values is more efficient that FFT based computation.
Example: DTMF Dual Tone Multifrequency
This is known as touch-tone/speed/electronic dialing
Pressing of each button generates a unique set of two-tone signals, called DTMF
signals.
These signals are processed at exchange to identify the number pressed by determining
the two associated tone frequencies.
Seven frequencies are used to code the 10 decimal digits and two special characters
(4x3 array)

 In this application frequency analysis requires determination of possible seven (eight)

DTMF fundamental tones and their respective second harmonics

For an 8 kHz sampling freq, the best value of the DFT length N to detect the eight
fundamental DTMF tones has been found to be 205
Not all 205 freq components are needed here, instead only those corresponding to key
frequencies are required
FFT algorithm is not effective and efficient in this application
The direct computation of the DFT can be formulated as a linear filtering operation on
the input data sequence.
This algorithm is known as Goertzel Algorithm
This algorithm exploits periodicity property of the phase factor
N 1

X (k ) = x(n)WNnk

(1)

n =0

SinceWN kN is equal to 1, multiplying both sides of the equation by this results in;
N 1

N 1

m =0

m =0

X (k ) = WN kN x(m)WNmk = x(m)WN k ( N m )

( 2)

yk (n) = x(n) hk (n)

This is in the form of a convolution

N 1

yk (n) = x(m)WN k ( n m )

(3)

m=0

hk (n) = WN knu (n)

( 4)

Where yk(n) is the out put of a filter which has impulse response of hk(n) and input
x(n)

 The output of the filter at n = N yields the value of the DFT at the freq k = 2k/N
The filter has frequency response given by
H k ( z) =

1
1 WN k z 1

( 6)

The above form of filter response shows it has a pole on the unit circle at the
frequency k = 2k/N
Entire DFT can be computed by passing the block of input data into a parallel bank of
N single-pole filters (resonators)
Difference Equation implementation of filter:
From the frequency response of the filter (eq 6) we can write the following difference
equation relating input and output;
Y ( z)
1
H k ( z) = k
=
X ( z ) 1 W N k z 1
y k (n) = W N k y k (n 1) + x(n)

y k (1) = 0

(7 )

The desired output is X(k) = yk(n) for k = 0,1,N-1

The phase factor appearing in the difference equation can be computed once and
stored.

The form shown in eq (7) requires complex multiplications which can be avoided
doing suitable modifications(divide and multiply by 1 W Nk z 1 )

The frequency response of the filter can be alternatively expressed as

1 WNk z 1
H k ( z) =
1 2 cos(2k / N ) z 1 + z 2

(8)

The direct form realization of the above is given by

vk (n) = 2 cos(2k / N )vk (n 1) vk (n 2) + x(n)
yk (n) = vk (n) WNk vk (n 1)

(9)

vk (1) = vk (2) = 0

(10)

 The recursive relation in (9) is iterated for

n = 0,1,N, but the equation in (10) is computed only once at time n =N
Each iteration requires one real multiplication and two additions.
Thus, for a real input sequence x(n) this algorithm requires (N+1) real multiplications
to yield X(k) and X(N-k) (this is due to symmetry)
Going through the Goertzel algorithm it is clear that this algorithm is useful only when
M out of N DFT values need to be computed where M 2log2N, Otherwise, the FFT
algorithm is more efficient method.

Chirp z- Transform
Situations where DFT computation through FFT is not preferable:
Computation of DFT is equivalent to samples of the z-transform of a finite-length
sequence at equally spaced points around the unit circle.
The efficient computation of DFT through FFT requires N to be a highly composite
number
Many a times we may need samples of z-transform on contours other than unit circle
or dense set of frequency samples over a small region of unit circle.
Consider these problems:
Obtain samples of z-transform on a circle of radius a which is concentric to unit
circle
Soln; The sequence need to be multiplied a-n
128 samples needed between frequencies
= -/8 to +/8 from a 128 point sequence
soln: 1024- point FFT computation where the given 128-point point sequence is
appended with 896 zeros. But only 128 frequencies out of 1024 needed, hence
wastage of computrations.

Chirp z-transform is the alternative.

Chirp z- transform is defined as:
N 1

X ( z k ) = x(n) z k n

k = 0,1,......L 1

(11)

n =0

Where zk is generalized contour.

Zk is the set of points in the z-plane falling on an arc which begins at some point z0
and spirals either in toward the origin or out away from the origin such that the
points {zk}are defined as,
z k = r0 e j 0 ( R0 e j0 ) k
k = 0,1,....L 1
(12)
Note that,
if R0< 1 the points fall on a contour that spirals toward the origin
If R0 > 1 the contour spirals away from the origin
If R0= 1 the contour is a circular arc of radius
If r0=1 and R0=1 the contour is an arc of the unit circle.
**Additionally this contour allows one to compute the freq content of the sequence
x(n) at a dense set of L frequencies in the range covered by the arc without having to
compute a large DFT (i.e., a DFT of the sequence x(n) padded with many zeros to
obtain the desired resolution in freq.)
If r0= R0=1 and 0=0 0=2/N and L = N the contour is the entire unit circle similar
to the standard DFT.

Substituting this value of zk in the expression of X(zk)

N 1

N 1

n =0

n =0

X ( z k ) = x(n) z k n = x(n)(r0 e j 0 ) nW nk

(13)

where
W = R0e j0

(14)

Expressing computation of X(zk) as linear filtering operation:

By substitution of
1
nk = (n 2 + k 2 (k n) 2 )
2
we can express X(zk) as

X ( zk ) = W k

/2

(15)

y (k ) = y (k ) / h(k )

k = 0,1,..........L 1

Where

h ( n) = W n

/2

g (n) = x(n)(r0 e j 0 ) n W n

/2

N 1

y ( k ) = g ( n ) h( k n)

(17)

n =0

both g(n) and h(n) are complex valued sequences

(16)

Why it is called Chirp z-transform?

if R0 =1, then seq h(n) has the form of complex exponential with argument
n = n20/2 = (n 0/2) n. The quantity (n 0/2) represents the freq of the complez
exponential signal, which increases linearly with time. Such signals are used in radar
systems are called chirp signals. Hence the name chirp z-transform.

Evaluation of linear convolution in eq (17)

Can be done efficiently with FFT

g(n) is finite length seq of length N and h(n) is of infinite duration, but fortunately
only a portion of h(n) is required to compute L values of X(z)
Since convolution is via FFT, it is circular convolution of the N-point seq g(n) with
an M- point section of h(n) where M > N
The concepts used in over lap save method can be used
While circular convolution is used to compute linear convolution of two sequences
we know the initial N-1 points contain aliasing and the remaining points are
identical to the result that would be obtained from a linear convolution of h(n) and
g(n)
In view of this the DFT size selected is M = L+N-1 which would yield L valid
points and N-1 points corrupted by aliasing.
The section of h(n) considered is for (N-1) n (L-1) yielding total length M as
defined
The portion of h(n) can be defined in many ways
h1(n) = h(n-N+1) n = 0,1,..M-1
Compute H1(k) and G(k) to obtain
Y1(k) = G(K)H1(k)
Application of IDFT will give y1(n), for
n =0,1,M-1

The starting N-1 are discarded and desired values are y1(n) for N-1 n M-1 which
corresponds to the range 0 n L-1 i.e.,
y(n)= y1(n+N-1) n=0,1,2,..L-1
Alternatively h2(n) can be defined as
h2 (n) = h(n)

0 n L 1

h(n ( N + L 1))

L n M 1

Compute Y2(k) = G(K)H2(k)

The desired values of y2(n) are in the range
0 n L-1 i.e.,
y(n) = y2(n) n=0,1,.L-1
Finally, the complex values X(zk) are computed by dividing y(k) by h(k)
For k =0,1,L-1

Computational complexity
In general the computational complexity of CZT is of the order of M log2M
complex multiplications
This should be compared with N.L which is required for direct evaluation
If L is small direct evaluation is more efficient otherwise if L is large then CZT is
more efficient.
Advantages of CZT
Not necessary to have N =L
Neither N or L need to be highly composite
The samples of Z transform are taken on a more general contour that includes the
unit circle as a special case.
Example to understand utility of CZT algortithm in freq analysis
(ref: DSP by Oppenheim Schaffer)
CZT is used in this application to sharpen the resonances by evaluating the ztransform off the unit circle.
Signal to be analyzed is a synthetic speech signal generated by exciting a five-pole
system with a periodic impulse train. The system was simulated to correspond to a
sampling frq of 10khz.
The poles are located at center freqs of 270,2290,3010,3500 & 4500 Hz with
bandwidth of 30, 50,60,87 & 140 Hz respectively.

The first two spectra correspond to spiral contours outside the unit circle with a
resulting broadening of the resonance peaks
|w| = 1 corresponds to evaluating z-transform on the unit circle
The last two choices corresponds to spiral contours which spirals inside the unit
circle and close to the pole locations resulting in a sharpening of resonance peaks.

Implementation of CZT in hardware to compute the DFT signals

DFT computation requires r0 =R0 =1, 0 = 0
0 = 2/N and L = N
The cosine and sine sequences in h(n) needed for pre multiplication and post
multiplication are usually stored in a ROM
If only magnitude is of DFT is desired, the post multiplication are unnecessary
In this case |X(zk)| = |y(k)| k =0,1,.N-1

Design of Digital Filters

Two important classes of filters based on impulse response type are
Finite Impulse Response (FIR)
Infinite Impulse Response (IIR)
The filter can be expressed as:
System function representation;
M

b z

H ( z) =

k =0
N

1 + ak z

(1)
k

k =1

Difference Equation representation;

N

k =0

k =0

ak y (n k ) = bk x(n k )

( 2)

Each of this form allows various methods of implementation

The eq (2) can be viewed as a computational procedure (an algorithm) for determining
the output sequence y(n) of the system from the input sequence x(n)
Different realizations possible with different arrangements of eq (2)

Filter Design Issues:

Realizability
Stability
Sharp Cutoff Characteristics
Minimum order
Generalized procedure
Linear phase characteristics

Issues considered for filter implementation:

Simple design
Structuredness modularity
Generalization of design any filter type one design method
Cost of implementation
Software/Hardware realization

Features of IIR:
Out put is a function of past o/p, present and past i/ps
Recursive nature
Has at least one Pole (in general poles and zeros)
Sharp cutoff chas. achievable with minimum order
Difficult to have linear phase chas over full range of freq.
Typical design procedure is analog design then conversion from analog to digital
Features of FIR ;

Inherently Stable
Linear phase characteristics possible
Simple implementation both recursive and nonrecursive structures possible
Free of limit cycle oscillations when implemented on a finite-word length digital
system
Disadvantages:
Sharp cutoff at the cost of higher order
Higher order leading to more delay, more memory and higher cost of implementation

Importance of Linear Phase:

The group delay is defined as
d ( )
g =
d
Nonlinear phase results in different frequencies experiencing different delay and
arriving at different time at the receiver
This creates problems with speech processing and data communication applications
Understanding simple FIR filtering operations:
1.Unity Gain Filter
y(n)=x(n)
2. Constant gain filter
y(n)=Kx(n)
3. Unit delay filter
y(n)=x(n-1)
4.Two - term Difference filter
y(n) = x(n)-x(n-1)
5. Two-term average filter
y(n) = 0.5(x(n)+x(n-1))
6. Three-term average filter (3-point moving average filter)
y(n) = 1/3[x(n)+x(n-1)+x(n-2)]
7. Central Difference filter
y(n)= 1/2[ x(n) x(n-2)]

Order of the filter is the number of previous inputs used to compute the current
output
Filter coefficients are the numbers associated with each of the terms x(n), x(n-1),.. Etc
The table below shows order and filter coefficients of above simple filter types:

Ex.

order

1
0
2
0
3
1
4(HP) 1
5(LP) 1
6(LP) 2
7(HP) 2

a0

a1

a2

1
K
0
1
1/2
1/3
1/2

1
-1
1/2
1/3
0

1/3
-1/2

Design of FIR filters:

Symmetric and Antisymmetric FIR filters giving out Linear Phase characteristics:
Symmetry in filter impulse response will ensure Linear phase
An FIR filter of length M with i/p x(n) & o/p y(n) is described by the difference equation:
M 1

y(n)= b0 x(n) + b1 x(n-1)+.+b M-1 x(n-(M-1)) =

b x(n k )
k

-(1)

k =0

Alternatively. it can be expressed in convolution form

M 1

y (n) =

h( k ) x( n k )

- (2)

k =0

i.e bk= h(k), k=0,1,..M-1

Filter is also characterized by
M 1

H ( z) =

h(k ) z

-(3) polynomial of degree M-1 in the variable z-1. The roots of this

k =0

polynomial constitute zeros of the filter.

An FIR filter has linear phase if its unit sample response satisfies the condition
h(n)= h(M-1-n) n=0,1,.M-1
-(4)

Incorporating this symmetry & anti symmetry condition in eq 3 we can show linear
phase chas of FIR filters
H ( z ) = h(0) + h(1) z 1 + h(2) z 2 + ........... + h( M 2) z ( M 2) + h( M 1) z ( M 1)
If M is odd

H ( z ) = h(0) + h(1) z 1 + .......... + h(

M 1 (
)z
2

M 1
)
2

+ h(

M + 1 (
)z
2

M +1
)
2

+ h(

M + 3 (
)z
2

M +3
)
2

+ ...........

+ h( M 2) z ( M 2) + h( M 1) z ( M 1)

M 1
M 3
M 1
(
)
(
)
(
)

M 1
M + 1 1
M + 3 2
2
2
=z
+ h(1) z
+ ............ + h(
) + h(
) z + h(
) z + .....h( M 1) z 2
h ( 0) z
2
2
2

Applying symmetry conditions for M odd

(

M 1
)
2

h(0) = h( M 1)
h(1) = h( M 2)
.
.
M 1
M 1
h(
) = h(
)
2
2
M +1
M 3
h(
) = h(
)
2
2
.
.
h( M 1) = h(0)
M 3

2
M
1

( M 1 2 n ) / 2
( M 1 2 n ) / 2

H ( z) = z
h(
) + h(n){z
z
}

2
n =0

similarly for M even

H ( z) = z

M 1
)
2

M 1
)
2

M2 1

h(n){z ( M 1 2 n ) / 2 z ( M 1 2 n ) / 2 }

n =0

Frequency response:
If the system impulse response has symmetry property (i.e.,h(n)=h(M-1-n)) and M is odd

H (e j ) = e j ( ) | H r (e j ) | where
M 3

2
M

1
M 1
j

) + 2 h(n) cos (
H r (e ) = h (
n)

2
2
n=0

M 1
( ) = (
) if | H r (e j ) | 0
2
M 1
= (
) + if | H r (e j ) | 0
2
In case of M even the phase response remains the same with magnitude response
expressed as

M2 1

M 1
j

H r (e ) = 2 h(n) cos (
n)
n=0

If the impulse response satisfies anti symmetry property (i.e., h(n)=-h(M-1-n))then for
M odd we will have
M 1
M 1
M 1
h(
) = h(
) i.e., h(
)=0
2
2
2
M23

M 1
j

H r (e ) = 2 h(n) sin (
n)
n =0

If M is even then,

M2 1

M 1
j

H r (e ) = 2 h(n) sin (
n)
n=0

In both cases the phase response is given by

M 1
) + / 2 if | H r (e j ) | 0
2
M 1
= (
) + 3 / 2 if | H r (e j ) | 0
2

( ) = (

Which clearly shows presence of Linear Phase characteristics.

Comments on filter coefficients:

The number of filter coefficients that specify the frequency response is (M+1)/2 when
is M odd and M/2 when M is even in case of symmetric conditions
In case of impulse response antisymmetric h(M-1/2)=0 so that there are (M-1/2) filter
coefficients when M is odd and M/2 coefficients when M is even

Choice of Symmetric and antisymmetric unit sample response

If h(n)=-h(M-1-n) and M is odd, Hr(w) implies that Hr(0)=0 & Hr()=0, consequently
not suited for lowpass and highpass filter.
Similarly if M is even Hr(0)=0 hence not used for low pass filter
Hence antisymetric condition is not generally used
Symmetry condition h(n)=h(M-1-n) yields a linear-phase FIR filter with non zero
response at w=0 if desired.

Zeros of Linear Phase FIR Filters

Consider the filter system function
M 1

H ( z ) = h ( n) z n
n =o

Expanding this equation

H ( z ) = h(0) + h(1) z 1 + h(2) z 2 + KK + h( M 2) z ( M 2) + h( M 1) z ( M 1)

sin ce for Linear phase we need
h(n) = h( M 1 n) i.e.,
h(0) = h( M 1); h(1) = h( M 2);......h( M 1) = h(0);
then
H ( z ) = h( M 1) + h( M 2) z 1 + ........ + h(1) z ( M 2) + h(0) z ( M 1)
H ( z ) = z ( M 1) [h( M 1) z ( M 1) + h( M 2) z ( M 2) + ..... + h(1) z + h(0)]
M 1

H ( z ) = z ( M 1) [ h(n)( z 1 ) n ] = z ( M 1) H ( z 1 )
n =0

This shows that if z = z1 is a zero then z=z1-1 is also a zero

The different possibilities:

If z1 = 1 then z1 = z1-1 =1 is also a zero implying it is one zero

If the zero is real and |z|<1 then we have pair of zeros

If zero is complex and |z|=1then and we again have pair of complex zeros.

If zero is complex and |z|1 then and we have two pairs of complex zeros

The plot above shows distribution of zeros for a Linear phase FIR filter.

Methods of designing FIR filters:

Fourier series based method
Window based method
Frequency sampling method
Design of Linear Phase FIR filter based on Fourier Series method:
Motivation: Since the desired freq response Hd(ej) is a periodic function in with
period 2, it can be expressed as Fourier series expansion
H d (e j ) =

h ( n) e

jn

n =

where hd (n) are

hd (n) =

1
2

fourier series coefficients

(e j )e jn d

This expansion results in impulse response coefficients which are infinite in duration and
non causal.
It can be made finite duration by truncating the infinite length

 The linear phase can be obtained by introducing symmetric property in the filter
impulse response, i.e., h(n) = h(-n)
It can be made causal by introducing sufficient delay (depends on filter length)
Stepwise procedure:

From the desired freq response using inverse FT relation obtain hd(n)

Truncate the infinite length of the impulse response to finite length with
assuming M odd)
h(n) = hd (n) for ( M 1) / 2 n ( M 1) / 2

= 0 otherwise
Introduce h(n) = h(-n) for linear phase characteristics
Write the expression for H(z); this is non-causal realization
To obtain causal realization H(z) = z -(M-1)/2 H(z)

Prob: Design an ideal bandpass filter with a frequency response:

H d (e j ) = 1
=0

for

4
otherwise

3
4

Find the values of h(n) for M = 11 and plot the frequency response.

1
hd (n) =
2

(e j )e jn d

/ 4
3 / 4

1
jn
jn
=
e d + e d
2 3 / 4
/4

1 3

=
sin
n sin n n

4
4
n
truncating to 11 samples we have h(n) = hd (n) for | n | 5

= 0 otherwise

 For n = 0 the value of h(n) is separately evaluated from the basic integration
h(0) = 0.5
Other values of h(n) are evaluated from h(n) expression
h(1)=h(-1)=0
h(2)=h(-2)=-0.3183
h(3)=h(-3)=0
h(4)=h(-4)=0
h(5)=h(-5)=0

The transfer function of the filter is

( N 1) / 2

H ( z ) = h ( 0) +

[h(n){z

+ z n }]

n =1

= 0.5 0.3183( z 2 + z 2 )
the transfer function of the realizable filter is
H ' ( z ) = z 5 [0.5 0.3183( z 2 + z 2 )]
= 0.3183 z 3 + 0.5 z 5 0.3183 z 7
the filter coeff are
h ' (0) = h' (10) = h' (1) = h' (9) = h' (2) = h' (8) = h' (4) = h' (6) = 0
h' (3) = h' (7) = 0.3183
h' (5) = 0.5

The magnitude response can be expressed as

( N 1) / 2
j

| H (e ) |=

a(n) cos n
n =1

comparing this exp with

5

| H (e j ) |=| z 5 [h(0) + 2 h(n) cos n] |

n =1

We have
a(0)=h(0)
a(1)=2h(1)=0
a(2)=2h(2)=-0.6366
a(3)=2h(3)=0
a(4)=2h(4)=0
a(5)=2h(5)=0
The magnitude response function is
|H(e j)| = 0.5 0.6366 cos 2 which can plotted for various values of
in degrees =[0 20 30 45 60 75 90 105 120 135 150 160 180];

|H(e j)| in dBs= [-17.3 -38.17 -14.8 -6.02 -1.74 0.4346 1.11 0.4346 -1.74 -6.02 -14.8 38.17 -17.3];

Prob: Design an ideal lowpass filter with a freq response

H d (e j ) = 1

for

2
Find the values of h(n) for N =11. Find H(z). Plot the magnitude response
=0

for

From the freq response we can determine hd(n),

/2

sin

1
2
e j n d =

2 / 2
n
Truncating hd(n) to 11 samples
hd ( n ) =

n and

n0

h(0) = 1/2
h(1)=h(-1)=0.3183
h(2)=h(-2)=0
h(3)=h(-3)=-0.106
h(4)=h(-4)=0
h(5)=h(-5)=0.06366
The realizable filter can be obtained by shifting h(n) by 5 samples to right h(n)=h(n-5)

h(n)= [0.06366, 0, -0.106, 0, 0.3183, 0.5, 0.3183, 0, -0.106, 0, 0.06366];

H ' ( z ) = 0.06366 0.106 z 2 + 0.3183 z 4 + 0.5 z 5 + 0.3183 z 6 0.106 z 8 + 0.06366 z 10
Using the result of magnitude response for M odd and symmetry
M 3
2

M 1
M 1
) + h(n) cos (
n)]
2
2
n =0
| H r (e j ) |=| [0.5 + 0.6366 cos w 0.212 cos 3w + 0.127 cos 5w] |
H r (e j ) = [h(

Exercise Problem:
Design an ideal band reject filter with a frequency response:
H d (e j ) = 1
=0

for

3
otherwise

and

2
3

Find the values of h(n) for M = 11 and plot the frequency response

Ans:h(n)= [0 -0.1378 0 0.2757 0 0.667 0 0.2757 0

-0.1378 0];

Window based Linear Phase FIR filter design

The arbitrary truncation of impulse response obtained through inverse Fourier relation
can lead to distortions in the final frequency response.
The arbitrary truncation is equivalent to multiplying infinite length function with finite
length rectangular window, i.e.,
h(n) = hd(n) w(n) where w(n) = 1 for n = (M-1)/2
The above multiplication in time domain corresponds to convolution in freq domain,
i.e.,
H( e j ) = Hd(e j) * W(e j ) where W(e j ) is the FT of window function w(n).

The FT of w(n) is given by

W (e j ) =

sin(M / 2)
sin( / 2)

 Suppose the filter to be designed is Low pass filter then the convolution of ideal filter
freq response and window function freq response results in distortion in the resultant
filter freq response. The ideal sharp cutoff chars are lost and presence of ringing effect
is seen at the band edges which is referred to Gibbs Phenomena.
This is due to main lobe width and side lobes of the window function freq response.

 The main lobe width introduces transition band and side lobes results in rippling
characters in pass band and stop band.
Smaller the main lobe width smaller will be the transition band
The ripples will be of low amplitude if the peak of the first side lobe is far below the
main lobe peak.
How to reduce the distortions?
Increase length of the window
- as M increases the main lob width becomes narrower, hence the transition band width
is decreased
-With increase in length the side lobe width is decreased but height of each side lobe
increases in such a manner that the area under each sidelobe remains invariant to
changes in M. Thus ripples and ringing effect in pass-band and stop-band are not
changed.
Choose windows which tapers off slowly rather than ending abruptly
- Slow tapering reduces ringing and ripples but generally increases transition width
since main lobe width of these kind of windows are larger.

What is ideal window characteristics?

Window having very small main lobe width with most of the energy contained with it
(i.e.,ideal window freq response must be impulsive)

Window design is a mathematical problem

More complex the window lesser are the distortions

Windows better than rectangular window are, Hamming, Hanning, Blackman, Bartlett,
Traingular,Kaiser

Rectangular window
wr (n) = 1 for 0 n M 1

Hanning windows:
whan (n) = 0.5(1 cos

2n
) for 0 n M 1
M 1

Hamming windows:
wham (n) = 0.54 0.46 cos

2n
for 0 n M 1
M 1

Blackman windows:
wblk (n) = 0.42 0.5 cos

2n
4n
+ 0.08 cos
for 0 n M 1
M 1
M 1

Bartlett (Triangular) windows:

M 1
|
2
M 1

2|n
wbart (n) = 1

for 0 n M 1

Kaiser windows:
2

2
M

1
M

I0
n

2
2

wk (n) =
M 1
I 0

for 0 n M 1

Type of window Appr. Transition

Peak
width of the main lobe sidelobe (dB)

Rectangular

4/M

-13

Bartlett

8/M

-27

Hanning

8/M

-32

Hamming

8/M

-43

Blackman

12/M

-58

Procedure for designing linear-phase FIR filters using windows

1. From the desired freq response using inverse FT relation obtain hd(n)
2. Truncate the infinite length of the impulse response to finite length with
( assuming M odd) choosing proper window
h(n) = hd (n) w(n) where
w(n) is the window function defined for ( M 1) / 2 n ( M 1) / 2
3.
4.
5.

Introduce h(n) = h(-n) for linear phase characteristics

Write the expression for H(z); this is non-causal realization
To obtain causal realization H(z) = z -(M-1)/2 H(z)

Prob: Design an ideal highpass filter with a frequency response:

H d (e j ) = 1
=0

for

| |<

4
using a hanning window with M = 11 and plot the frequency response.

1
hd ( n ) =
[
2

/ 4

j n

d +

j n

d ]

/4

1
n
[sin n sin ] for n and
4
n
/ 4

1
3
hd ( 0) =
[ d + d ] = = 0.75
2
4
/4
hd ( n ) =

hd(1) = hd(-1)=-0.225
hd(2) = hd(-2)= -0.159
hd(3) = hd(-3)= -0.075
hd(4) = hd(-4)= 0
hd(5) = hd(-5) = 0.045
The hamming window function is given by

2n
M 1
otherwise

whn (n) = 0.5 + 0.5 cos

=0
for

M 1
M 1
)n(
)
2
2

N = 11

whn (n) = 0.5 + 0.5 cos

n
5

whn(0) = 1
whn(1) = whn(-1)=0.9045
whn(2)= whn(-2)=0.655
whn(3)= whn(-3)= 0.345
whn(4)= whn(-4)=0.0945
whn(5)= whn(-5)=0

5 n 5

n0

h(n)= whn(n)hd(n)
h(n)=[0 0 -0.026 -0.104 -0.204 0.75 -0.204 -0.104 -0.026 0 0]

h' (n) = h(n 5)

H ' ( z ) = 0.026 z 2 0.104 z 3 0.204 z 4 + 0.75 z 5 0.204 z 6 0.104 z 7 0.026 z 8
Using the equation
M 3

2
M 1
M 1
H r (e jw ) = [h(
) + 2 h(n) cos (
n)
2
2
n =0
4

H r (e jw ) = 0.75) + 2 h(n) cos (5 n)

n=0

The magnitude response is given by,

|Hr(e j)| = |0.75 - 0.408cos - 0.208 cos2 - 0.052cos3|

in degrees = [0 15 30 45 60 75 90 105 120 135 150 165 180]

|H(e j)| in dBs = [-21.72 -17.14 -10.67 -6.05 -3.07 -1.297 -0.3726
-0.0087 0.052 0.015 0 0 0.017]

Prob: Design a filter with a frequency response:

H d (e j ) = e j 3
=0

for

<| |

using a Hanning window with M = 7

Soln:
The freq resp is having a term e j(M-1)/2 which gives h(n) symmetrical about

n = M-1/2 = 3 i.e we get a causal sequence

hd (n) =

1
2

/4

j 3

e j n d

/4

(n 3)
4
(n 3)
this gives hd (0) = hd (6) = 0.075
sin

hd (1) = hd (5) = 0.159

hd (2) = hd (4) = 0.22
hd (3) = 0.25

The Hanning window function values are given by

whn(0) = whn(6) =0
whn(1)= whn(5) =0.25
whn(2)= whn(4) =0.75
whn(3)=1
h(n)=hd(n) whn(n)
h(n)=[0 0.03975 0.165 0.25 0.165 0.3975 0]

Design of Linear Phase FIR filters using Frequency Sampling

method
Motivation: We know that DFT of a finite duration DT sequence is obtained by sampling
FT of the sequence then DFT samples can be used in reconstructing original time

domain samples if frequency domain sampling was done correctly. The samples of FT
of h(n) i.e., H(k) are sufficient to recover h(n).
Since the designed filter has to be realizable then h(n) has to be real, hence even
symmetry properties for mag response |H(k)| and odd symmetry properties for phase
response can be applied. Also, symmetry for h(n) is applied to obtain linear phase
chas.
Fro DFT relationship we have
h ( n) =

1
N

N 1

H ( k )e

j 2kn / N

n = 0,1,......N 1

for

k =0

N 1

H (k ) = h(n)e j 2kn / N

for

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

n=0

Also we know H(k) = H(z)|z=e j2kn/N

The system function H(z) is given by
N 1

H ( z ) = h( n ) z n
n =0

Substituting for h(n) from IDFT relationship

H ( z) =

1 z N
N

N 1

1 e

H (k )
j 2kn / N

k =0

z 1

Since H(k) is obtained by sampling H(ej) hence the method is called Frequency
Sampling Technique.
Since the impulse response samples or coefficients of the filter has to be real for filter to
be realizable with simple arithmetic operations, properties of DFT of real sequence can
be used. The following properties of DFT for real sequences are useful:
H*(k) = H(N-k)
|H(k)|=|H(N-k)| - magnitude response is even

(k) = - (N-k) Phase response is odd

h ( n) =

1
N

N 1

H ( k )e
k =0

j 2kn / N

can be rewritten as (for N odd)

N 1

H
(
0
)
H (k )e j 2kn / N
+

k =1

N 1 / 2
N 1
1

h(n) = H (0) + H (k )e j 2kn / N + H (k )e j 2kn / N

N
k =1
k = N 1 / 2

h ( n) =

1
N

Using substitution k = N r or r = N- k in the second substitution

with r going from now (N- 1)/2 to 1 as k goes from 1 to (N-1)/2
h ( n) =

1
N

( N 1) / 2
( N 1) / 2

j 2kn / N
H
(
0
)
+
H
(
k
)
e
+
H ( N k )e j 2kn / N

k =1
k =1

h ( n) =

1
N

( N 1) / 2
( N 1) / 2

j 2kn / N
H
(
0
)
+
H
(
k
)
e
+
H * (k )e j 2kn / N

k =1
k =1

h ( n) =

1
N

( N 1) / 2
( N 1) / 2

j 2kn / N
H
(
0
)
+
H
(
k
)
e
+
( H (k )e j 2kn / N ) *

k =1
k =1

h ( n) =

1
N

( N 1) / 2

H
(
0
)
+
( H (k )e j 2kn / N + ( H (k )e j 2kn / N ) *

k =1

h ( n) =

1
N

( N 1) / 2

H
(
0
)
+
2
Re( H (k )e j 2kn / N

k =1

Similarly for N even we have

( N 1) / 2

1
H
(
0
)
+
2
Re( H (k )e j 2kn / N

N
k =1

Using the symmetry property h(n)= h (N-1-n) we can obtain Linear phase FIR filters
using the frequency sampling technique.

h ( n) =

Prob: Design a LP FIR filter using Freq sampling technique having cutoff freq of /2
rad/sample. The filter should have linear phase and length of 17.

The desired response can be expressed as

j

H d (e ) = e
=0

j (

M 1
)
2

for | | c

otherwise

with M = 17 and c = / 2

H d (e j ) = e j 8

for 0 / 2
for / 2

=0

Selecting k =

2k 2k
=
M
17

H (k ) = H d (e j ) |
2k
8
17

for

k = 0,1,......16

2k
17

2k

17
2
2k
/2
=0
for

17
16k
j
17
H (k ) = e 17
for 0 k
4
17
17
=0
for
k
4
2
H (k ) = e

for 0

The range for k can be adjusted to be an integer such as

0k 4
and

5k 8

The freq response is given by

H (k ) = e
=0

2k
8
17

for

for 0 k 4
5k 8

Using these value of H(k) we obtain h(n) from the equation

h ( n) =

( M 1) / 2
1
( H (0) + 2 Re( H (k )e j 2kn / M ))
M
k =1

i.e.,
4
1
(1 + 2 Re(e j16k / 17 e j 2kn / 17 ))
17
k =1
4
1
2k (8 n)
h(n) = ( H (0) + 2 cos(
) for
17
17
k =1

h ( n) =

n = 0,1,........16

Even though k varies from 0 to 16 since we considered varying between 0 and

/2 only k values from 0 to 8 are considered
While finding h(n) we observe symmetry in h(n) such that n varying 0 to 7 and 9
to 16 have same set of h(n)

Design of FIR Differentiator

Differentiator are widely used in Digital and Analog systems whenever a derivative of
the signal is needed

Ideal differentiator has pure linear magnitude response in the freq range to +

Prob: Design an Ideal Differentiator using a) rectangular window and b)Hamming

window with length of the system = 7.
Solution:
As seen from differentiator frequency chars. It is defined as
H(ej) = j

between to +

1
cos n
j e j n d =
n and n 0

2
n
The hd(n) is an add function with hd(n)=-hd(-n) and hd(0)=0
hd ( n ) =

a) rectangular window
h(n)=hd(n)wr(n)
h(1)=-h(-1)=hd(1)=-1

h(2)=-h(-2)=hd(2)=0.5
h(3)=-h(-3)=hd(3)=-0.33
h(n)=h(n-3) for causal system
thus,
H ' ( z ) = 0.33 0.5 z 1 + z 2 z 4 + 0.5 z 5 0.33 z 6
Also from the equation
( M 3) / 2

H r (e j ) = 2

h(n) sin (

n =0

M 1
n)
2

For M=7 and h(n) as found above we obtain this as

H r (e j ) = 0.66 sin 3 sin 2 + 2 sin
H (e j ) = jH r (e j ) = j (0.66 sin 3 sin 2 + 2 sin )

b) Hamming window
h(n)=hd(n)wh(n)
where wh(n) is given by
wh (n) = 0.54 + 0.46 cos

2n
( M 1)

( M 1) / 2 n ( M 1) / 2

= 0 otherwise
For the present problem
wh (n) = 0.54 + 0.46 cos

3 n 3
3
The window function coefficients are given by for n=-3 to +3
Wh(n)= [0.08 0.31 0.77 1 0.77 0.31 0.08]
Thus h(n) = h(n-5) = [0.0267, -0.155, 0.77, 0, -0.77, 0.155, -0.0267]
Similar to the earlier case of rectangular window we can write the freq response of
differentiator as
H (e j ) = jH r (e j ) = j (0.0534 sin 3 0.31sin 2 + 1.54 sin )

We observe
With rectangular window, the effect of ripple is more and transition band width is
small compared with hamming window
With hamming window, effect of ripple is less whereas transition band is more

Design of FIR Hilbert transformer

Hilbert transformers are used to obtain phase shift of 90 degree

They are also called j operators
These signals are needed in quadrature signal processing
The Hilbert transformer is very useful when out of phase component (or imaginary
part) need to be generated from available real component of the signal.

Design an ideal Hilbert transformer using a) rectangular window and b) Blackman

Window with M = 11

Solution:
As seen from freq chars it is defined as
H d (e j ) = j 0
0

=j

The impulse response is given by

0

1
(1 cos n)
[ je jn d + je jn d ] =
n except n = 0
2
n
0
At n = 0 it is hd(0) = 0 and hd(n) is an odd function
hd ( n ) =

a) Rectangular window
h(n) = hd(n) wr(n) = hd(n) for -5 n 5
h(n)=h(n-5)
h(n)= [-0.127, 0, -0.212, 0, -0.636, 0, 0.636, 0, 0.212, 0, 0.127]
4

H r (e j ) = 2 h(n) sin (5 n)
n =0
j

H (e ) = j | H r (e j ) |= j{0.254 sin 5 + 0.424 sin 3 + 1.272 sin }

b) Blackman Window
window function is defined as
n
2n
wb (n) = 0.42 + 0.5 cos + 0.08 cos
5
5
=0
otherwise

5 n 5

Wb(n) = [0, 0.04, 0.2, 0.509,0.849,1,0.849, 0.509, 0.2, 0.04,0] for -5n5
h(n) = h(n-5) = [0, 0, -0.0424, 0, -0.5405, 0, 0.5405, 0, 0.0424, 0, 0]
H (e j ) = j[0.0848 sin 3 + 1.0810 sin ]

Frequency Transformation
Why Frequency Transformation:
This allows one to design prototype filter and transform to any specific frequency
selective type
Designers can concentrate on improved methods of designing prototype rather than
wasting time on devising design methodologies for different types of filters
One design and all types of frequency selective filters is an advantage

Where Frequency Transformation

Analog Domain
Digital Domain
Transformation in Digital Domain
Frequency Transformation in Analog domain:
Normalized Low Pass filter with cutoff freq of p = 1 rad/sec is designed
Normalized LPF to LPF of specific cutoff:

s
'p

H1 ( s) = H p (

p
'p

s)

Where,
p= normalized cutoff freq=1 rad/sec
p= Desired LP cutoff freq
at =p it is H(j1)

Frequency Transformation in Digital Domain

Involves replacing the variable Z-1 by a rational function g(z-1)
Should satisfy the following properties .

Mapping Z-1 to g(z-1) must map points inside the unit circle in the Z- plane onto the
unit circle of z- plane .

For stable filter, the inside of the unit circle of the Z - plane must map onto the inside
of the unit circle of the z-plane.
The general form of the function g(.) that satisfy the above requirements of " all-pass "
type is

Prob:
Let

1.
2.
3.
4.
5.

H ( s) =

1
s2 + s +1

Represents the transfer function of a lowpass filter (not butterworth) with a passband
of 1 rad/sec. Use freq transformation to find the transfer function of the following
filters:
A LP filter with a passband of 10 rad/sec
A HP filter with a cutoff freq of 1 rad/sec
A HP filter with a cutoff freq of 10 rad/sec
A BP filter with a passband of 10 rad/sec and a corner freq of 100 rad/sec
A BS filter with a stopband of 2 rad/sec and a center freq of 10 rad/sec

Solution:
Given
H ( s) =

1
s + s +1
2

a. LP LP Transform
replace
s
s
s
=
p 10
sub H a ( s ) = H ( s ) |

100
s + 10 s + 100
2

s
10

1
s
s 2

( ) + ( ) + 1
10
10

b. LP HP(normalized) Transform

u 1
=
s
s

sub H a ( s ) = H ( s ) |

1
s
s

1
1 2 1

( ) + ( ) + 1
s
s

s2
s2 + s +1

c. LP HP(specified cutoff) Transform

replace

u 10
=
s
s

sub H a ( s ) = H ( s ) |

10
s
s

1
10 2 10

( ) + ( ) + 1
s
s

s2
= 2
s + 10 s + 100
c. LP BP Transform
replace
s

s 2 + u l s 2 + o2
=
s ( u l )
sB0

where o = u l

Bo = ( u l )

and

sub H a ( s ) = H ( s ) |
=

s 2 +10 4
10 s

100 s 2
s 4 + 10 s 3 + 20100 s 2 + 105 s + 108

c. LP BS Transform
replace
s ( l )
sB
s 2 u
= 2 0 2
s + u l s + o

where o = u l

Bo = ( u l )

and

sub H a ( s ) = H ( s ) |
=

2s
s 2 +100

( s 2 + 100) 2
s 4 + 2 s 3 + 204 s 2 + 200 s + 10 4

Prob:
Convert single pole LP Bufferworth filter with system function

Into BPF with upper & lower cutoff frequency

respectively ,
The LPF has 3-dB bandwidth
Soln

Note that the resulting filter has zeros at z=1 and a pair of poles that depend on the
choice of l and u

This filter has poles at z=j0.713 and hence resonates at =/2

Conclusion
It is shown here that how easy to convert one form of filter design to another
form.
What we require is only prototype low pass filter design steps to transform to
any other form.