Академический Документы
Профессиональный Документы
Культура Документы
Sr No TOPIC [ PAGE
1 Basic Concepts 2
5 Digital Filters 26
When you study DSP, so many questions come in mind. The more you study
more questions come. Those question needs to be resolved, I have made
the list of few FAQ questions. After you study DSP, you should be in a
position to answer these questions.
Ans : Digital Signal Processing is a technique that converts signals from real world sources
(usually in analog form) into digital data that can then be analyzed. Analysis is performed
in digital form because once a signal has been reduced to numbers, its components can be
isolated, analyzed and rearranged more easily than in analog form.
Eventually, when the DSP has finished its work, the digital data can be turned back into an
analog signal, with improved quality. For example, a DSP can filter noise from a signal,
remove interference, amplify frequencies and suppress others, encrypt information, or
analyze a complex waveform into its spectral components. This process must be handled in
real-time - which is often very quickly. For instance, stereo equipment handles sound
signals of up to 20 kilohertz (20,000 cycles per second), requiring a DSP to perform
hundreds of millions of operations per second.
2
Kiran Talele ( talelesir@yahoo.com 9987030881 )
(5) What do you know about Analog Signal, Digital Signal, CT signal, DT Signal ?
Ans : i] Analog Signal : Signal value can be anything. NO fixed signal level.
Eg x(t) =cos(100π t). Í continuous Sinusoidal signal
x[n] = { 10.5, 4.7, 3.5, 5.7, 3.8 } Í sampled signal
ii] Digital Signal : Only two levels +5v and 0. ie. Logically High and Low.
Eg. Binary data
iii] Continuous Time Signal : Signal is defined for every value of time. Signal value can be
anything.
Eg. x(t) =cos(100π t). Í continuous Sinusoidal signal
Bilevel Signal
iv] Discrete Time Signal : Signal is defined for Discrete instant of Time. NOT for every
value of time. Signal value can be anything.
x ( t ) t = nTs = x [ nTs ] = x [ n ]
Ans : When processing the analog signal using DSP System, it is sampled at some rate depending
upon the bandwidth. The rate of sampling is decided by the Nyquist criterion. However,
signals that are found in physical systems will never be strictly bandlimited. To eliminate
signal content beyond the desired bandwidth, antialiasing filter is used.
The filter cannot be a digital filter. This is because antialias filtering is required to be
performed in the analog domain prior to applying the signal to A/D converter where
aliasing would take place.
(8) Let x[t] = 10 cos(100π) + 20 cos( 120πt)-5 sin(50πt). If x(t) is sampled with sampling
frequency Fs = 200 Hz. What will be Discrete Time Signal x[n] at n=0 ?
Ans : 30
3
Kiran Talele ( talelesir@yahoo.com 9987030881 )
2. DISCRETE TIME SIGNALS
(10) What do you mean by Causal signal, Anti-causal Signal and Both-sided signal ?
Ans : Examples :
(i) Causal signal : x1[n] = u[n] x2[n] =( ½ )n u[n]
(ii) Anti-causal signal : x1[n] = u[-n-1] x2[n] =( ½ )n u[–n–1]
(iii) Both sided signal : x1[n] = u[n] + u[-n-1] x2[n] = (2)n u[n] + (3)n u[–n–1]
N −1
Ans : Energy of signal is defined as, E= ∑
n =0
x[n]
2
{ }
Ex : x[n] = ( ½ )n u[n] E = 2 ( finite)
x [n]= 1 2 3 4 E = 30 (finite)
↑
(13) Consider x1[n] is periodic with period = 4 and x2[n] is periodic with period = 6 .
Let x[n] = x1[n] + x2[n]. What will be the period of x[n] ?
4
Kiran Talele ( talelesir@yahoo.com 9987030881 )
(14) What is power signal ? Give example.
N −1
1
Ans : The average power of the x[n] is given as P = Nlim
→∞ N
∑ x[n]
n =0
2
⎧ ⎫ Periodic
x [n ] = ⎨ − 1 −2 3 3 −2 ⎬
⎩ ↑ ⎭
⎧ ⎫ Periodic
x p [n ] = ⎨ 0 − 2 3 − 3 2 ⎬
⎩ ↑ ⎭
⎛ −1 1 ⎤
Range of Digital frequency f is ⎜⎜ , ⎥
⎝ 2 2 ⎦
(23) What is the unit of digital frequency w and f ?
Ans : Unit of digital frequency w is radians and f is unit less quantity.
6
Kiran Talele ( talelesir@yahoo.com 9987030881 )
(29) Consider x1[n] is periodic with period N1 = 4 and x2[n] is periodic with period
N2 = 6 . Let x[n] = x1[n] + x2[n]. What will be the period of x[n] ?
Ans : Period N = LCM { N1, N2 } = 12
(30) Let x[n] = δ[n] + 2 u[n] – 2 u[n-4] . Determine which of the following classification is true for x[n].
(a) Periodic, Finite length (b) Periodic, Infinite length
(c) Non periodic, Finite length (d) Non-periodic, Infinite length
NOTE :
¾ Linear Shifting of NON-Periodic DT Signals
1) x[ n ] = ⎧ ⎫
⎨ 1 2 3 4 ⎬
⎩ ↑ ⎭
⎧ ⎫
2) x[ n –1 ] = ⎨ 0 1 2 3 4 ⎬
⎩ ↑ ⎭
⎧ ⎫
3) x[ n + 1 ] = ⎨ 1 2 3 4 ⎬
⎩ ↑ ⎭
⎧⎪ ⎫⎪
4) x[ –n ] = ⎨ 4 3 2 1 ⎬
⎪⎩ ↑ ⎪⎭
⎧⎪ ⎫⎪
5) x[ –n + 1 ] = ⎨ 4 3 2 1 ⎬
⎪⎩ ↑ ⎪⎭
⎧ ⎫
6) x[ –n – 1 ] = ⎪ ⎪
⎨ 4 3 2 1 0 ⎬
⎪ ↑ ⎪
⎩ ⎭
----------------------------------------------------------------------------------------------------
¾ Circular Shifting of Periodic DT Signals
1) x[ n ] = { 1, 2, 3, 4 }
2) x[ n-1 ] = { 4, 1, 2, 3 }
3) x[ n+1 ] = { 2, 3, 4, 1 }
4) x[ –n ] = { 1, 4, 3, 2 }
5) x[ –n+1 ] = { 2, 1, 4, 3 }
6) x[ –n–1 ] = { 4, 3, 2, 1 }
-----------------------------------------------------------------------------------------------------------
7
Kiran Talele ( talelesir@yahoo.com 9987030881 )
3. DFT and FFT
(34) How many real multiplications and additions are required to find DFT.
Ans :
Let P = a + j b and Q = c + j d
(35) How many real multiplications and additions are required to find DFT of 32 point signal.?
Ans : By DFT
(i) Real Multiplications = 4 N 2 = 4(32) 2 = 4096
(i) Real Additions = 4 N 2 − 2 N = 40321
8
Kiran Talele ( talelesir@yahoo.com 9987030881 )
(36) How many complex multiplications and additions are required to find FFT ?
Ans : By DFT
N
(i) Complex Multiplications = log 2 N
2
(ii) Complex Additions = N log 2 N
(37) How many real multiplications and additions are required to find DFT of 32 point signal using FFT
algorithm?
Ans : By FFT
(i) Real Multiplications = 2 N log 2 N = 320
(ii) Real Additions = 3 N log 2 N = 480
Linearity Property : If signals are added, Then DFT’s are also added.
i.e. DFT { a x 1 [n ] + b x 2 [n ] } = a X 1 [k ] + b X 2 [k
{ }
DFT W N− mn x [ n] = X [ k − m]
10
Kiran Talele ( talelesir@yahoo.com 9987030881 )
(55) What is the DFT of Imaginary and Odd signal ?
Ans : If x[n] is Imaginary and Odd
Then X[k] is also Real and Odd
Eg. x[n]={ 0, 2j, 0, – 2j }
X[k] = { 0, 4, 0, – 4 }
(56) If DT signal is expanded in time domain what will be the effect in frequency domain?
Ans : Expansion in time domain corresponds to Compression in frequency domain.
Eg. x[n] = {1,2,3,2 } X[k] = { 8, –2, 0, –2}
Let p[n] = {1, 0, 2, 0, 3, 0, 2,0 } Then P[k] = { 8, –2, 0, –2, 8, –2, 0, –2}
|X[k]| |P[k]|
(57) If DT signal is compressed in time domain what will be the effect in frequency domain?
Ans : Compression in time domain corresponds to Expansion in frequency domain.
Eg. x[n] = {1, 0, 2, 0, 3, 0, 2,0 } X[k] = { 8, –2, 0, –2, 8, –2, 0, –2}
Let p[n] = {1,2,3,2 } Then P[k] = { 8, –2, 0, –2 }
|X[k] |P[k]|
|
|X[k]| |P[k]|
Also, "mixed radix" FFT's also can be done on "composite" sizes. In this case, you break a
non-prime size down into its prime factors, and do an FFT whose stages use those
factors. For example, an FFT of siz 1000 might be done in six stages using radices of 2
and 5, since 1000 = 2 * 2 * 2 * 5 * 5 * 5. It might also be done in three stages using radix
10, since 1000 = 10 * 10 * 10.
Ans : An "in place" FFT is simply an FFT that is calculated entirely inside its original sample
memory. In other words, calculating an "in place" FFT does not require additional buffer
memory (as some FFT's do.)
x[n] DFT/FFT
X[k]
Y[k] y[n]
× iDFT /
h[n] DFT/FFT
H[k]
15
Kiran Talele ( talelesir@yahoo.com 9987030881 )
(80) How to find output of the IIR filter using DFT / FFT?
Ans : Output of the filter is Linear convolution of impulse response with the input of the
signal.
To find output means to find LC by DFT/FFT. Length of h[n] in IIR filter is infinite.
So, DFT/FFT implementation of infinite length signals is not possible.
(83) What is the difference between circular convolution and periodic convolution ?
Ans : In periodic convolution input signals are originally periodic with common value of period.
In circular convolution, if input signals are not periodic then they are assumed to be periodic
with period = N where N = max(L,M) where L is the length of first signal and M is length of
second signal.
(87) How to find output of FIR filter for long input sequence.
Ans : In FIR filter length of h[n] is finite. Output of the filter is always Linear Convolution of impulse
response with the input of the signal. To find output of digital FIR filter FFT technique is used. But
for Long data sequence, direct FFT technique is not suitable.
For long data sequence, Overlap Add Method using FFT or Overlap Save Method using FFT is
used.
(91) How to find output of IIR filter for real time input signal.?
Ans : In real time application entire input is not available and input signal has to be processed online.
Length of input signal depends on application. It can be long sequence also.
In IIR filter length of h[n] is infinite. Output of the filter is always Linear Convolution of impulse
response with the input of the signal. To find output of digital IIR filter, Overlap Add Method
using FFT or Overlap Save Method using FFT can not be used.
Output of digital IIR filter is calculated using difference equation recursively.
(92) How to find output of IIR filter for long input sequence.?
Ans : In IIR filter length of h[n] is infinite. Output of the filter is always Linear Convolution of impulse
response with the input of the signal. To find output of digital IIR filter, Overlap Add Method
using FFT or Overlap Save Method using FFT can not be used.
Output of digital IIR filter is calculated using difference equation recursively.
17
Kiran Talele ( talelesir@yahoo.com 9987030881 )
(98) How to find DFT of infinite length sequence ?
Ans : To find DFT of infinite length sequence x[n]:
∞
(i) Find DTFT of x[n] i.e. X ( w) = ∑ x[n] e − jnw
n = −∞
18
Kiran Talele ( talelesir@yahoo.com 9987030881 )
(102) What is the necessary condition to find DTFT of any signal. ?
Ans : To find DTFT of any signal the necessary condition is, signal must be an energy
signal. It must be absolutely summable.
(105) What is the difference between DFT and DTFS ? [Refer Notes]
(106) What is the relation between DFT and DTFS ? [Refer Notes]
(107) What is the relation between DFT and DTFT ? [Refer Notes]
(108) What is the relation between DTFT and ZT ? [Refer Notes]
(109) What is the relation between DFT and ZT ? [Refer Notes]
(110) How to find DFT of Two N point Real Sequence using a single N point FFT ?
(111) How to find DFT of 2N point DFT of real valued sequence using a single N point FFT
algorithm? [Refer Notes]
---------------------------------------------------------------------
19
Kiran Talele ( talelesir@yahoo.com 9987030881 )
4. ANALYSIS OF DT SYSTEM
NOTE : If x[n] is right handed sequence, the ROC extends outward from the
outermost finite pole in X ( z ) to z = ∞
Sequence ROC
1 x[n] = Entire Z-plane
NOTE : If x[n] is Left handed sequence, the ROC extends inward from the
innermost finite pole in X (z) to z = 0
Sequence ROC
1 x[n] = { 1, 2, 3, 0 } |Z| < ∞
(119) What is the ROC condition for Both-sided signal. ? Why ? Justify
with example.
Ans : ROC condition for both sided signal is bounded between two POLES.
Ex x[n] = (2)n u[n] + (3)n u[-n]
NOTE : If x[n] is two sided sequence, the ROC consist of a ring in the Z plane,
bounded by interior and exterior pole.]
Sequence ROC
1 x[n] = an u[n] + bn u[-n-1] |b| > |z| > |a|
2 x[n] = (2)n u[n] + (3)n u[-n-1] 3 > |z| > 2
22
Kiran Talele ( talelesir@yahoo.com 9987030881 )
(126) What is zero state step response ?
Ans : If the initial state of the system is zero and the input x[n]=u[n] then the output of the
system is called zero step response of the system.
2 2
Where Magnitude = (Real) + (Imaginary)
Phase Response = Summation of angles from ZEROS – Summation of angles from POLES.
23
Kiran Talele ( talelesir@yahoo.com 9987030881 )
(135) Magnitude spectrum is continuous or discrete ?
Ans : If the signal is periodic then magnitude spectrum is discrete and If the signal is not-
periodic then spectrum is continuous function of w.
24
Kiran Talele ( talelesir@yahoo.com 9987030881 )
(145) What is the difference equation of DT sy stem ?
Ans : Output in terms of past/present , input/output of the system is called difference of the
system.
Eg y[n] = y[n–1] + y[n–1] + x[n] + x[n–1]
NOTE : If the system is Neither Minimum Phase NOR Maximum Phase Then System
is Mixed Phase System.
Minimum Phase characteristic implies a min. delay function while a maximum
phase characteristic implies that the delay characteristic is also maximum.
25
Kiran Talele ( talelesir@yahoo.com 9987030881 )
Always remember this →
(i) h[n]
x[n] (ii) H[z] y[n]
(iii) Difference Equation
(iv) Realization Diagram
ZT (v) Pole Zero Plot IZT
X(z) Y(z)
H(z)
(i) Take ZT
P.Z.
(ii) Group the terms
with Y(z) & X(z)
(iii) Arrange in terms
of Y(z)/X(z)
IZT
D.E. H(z) h[n]
ZT
Put z = ejw
(153) Impulse response of Digital Low Pass filter is given by h[n] ={ 3, 2, 1, 2, 3 }. What will
be the output of the filter for any given input x[n] ?
x[n] y[n]
Digital Filter
26
Kiran Talele ( talelesir@yahoo.com 9987030881 )
5. DIGITAL FILTERS
2. Digital filters are easily designed, tested and implemented on a general-purpose computer
or workstation.
3. The characteristics of analog filter circuits (particularly those containing active
components) are subject to drift and are dependent on temperature. Digital filters do not
suffer from these problems, and so are extremely stable with respect both to time and
temperature.
27
Kiran Talele ( talelesir@yahoo.com 9987030881 )
(159) What are Advantages of FIR Filters-? [Refer Theory Notes ]
Ans:
(160) What are the disadvantages of FIR Filters (compared to IIR filters)?
Ans : Compared to IIR filters, FIR filters sometimes have the disadvantage that they require more
memory and/or calculation to achieve a given filter response characteristic.
(161) What are the advantages of IIR filters (compared to FIR filters)?
Ans : IIR filters can achieve a given filtering characteristic using less memory and calculations than a
similar FIR filter.
(162) What are the disadvantages of IIR filters (compared to FIR filters)?
Ans : 1) They are more susceptible to problems of finite-length arithmetic, such as noise generated by
calculations, and limit cycles. (This is a direct consequence of feedback: when the output
isn't computed perfectly and is fed back, the imperfection can compound.)
(164) What is the relation between Analog filter pole and digital filter pole when impulse invariant
technique is used for filter design.
Ans : Z = e ST
(165) What is the relationship between Analog filter frequency and digital filter frequency when
impulse invariant technique is used for filter design.
Ans : W = ΩT
28
Kiran Talele ( talelesir@yahoo.com 9987030881 )
(166) Why Impulse Invariant method is not suitable for HPF / BPF design?
Ans : The the mapping from the analog frequency Ω to the freq. variable w in the digital domain is
many to one. which reflects the effect of aliasing due to sampling. A one to one mapping is
π π
thus possible only if freq. Ω lies in the principle range of − ≤ Ω ≤ .
T T
π
That means if cut off frequency of analog filter Ω c is greater than . then one to one
T
mapping from analog filter frequency to digital filter frequency is not possible. Therefore the
π
filter such as HPF or BPF with cut off frequency of analog filter Ω c greater than .
T
can not be designed using impulse invariant method.
(168) Explain the Mapping of points from s-plane to z–plane when Impulse Invariant Method is used
for filter design.
Case-I When σ = 0, r = 1
Analog poles which lies on imaginary axis gets mapped onto the unit circle in
the z-plane.
(169) What is the relation between Analog filter pole and digital filter pole when BLT method is used
for filter design.
2 ( z − 1)
Ans : S =
T ( z + 1)
(170) What is the relationship between Analog filter frequency and digital filter frequency when
BLT method is used for filter design.
2 ⎛ w⎞
Ans : Ω= tan⎜ ⎟
T ⎝2⎠
(171) Explain frequency warping in BLT.
Ans : [ Refer theory notes ]
29
Kiran Talele ( talelesir@yahoo.com 9987030881 )
(172) Frequency warping is needed to perform in BLT technique but not in impulse
invariance techniqueOR In BLT there is no aliasing
Ans :
Bilinear Transformation is a mapping of points from s-plane to corresponding points
in the z-plane. The BLT transforms, the entire j Ω axis in the s-plane into one
revolution of the unit circle in the z-plane ie. only once and therefore avoids the aliasing of
frequency components.
[A] For Linear Phase filter h[n] must be either Symmetric or Antisymmetric.
[B] When h[n] is either Symmetric OR Antisymmetric, ZEROS of the filter are always
in Reciprocal order.
1
i.e. If Z1 is ZERO of the filter, Then is also a ZERO of the filter.
z1
[C] If ZEROS of the filter are in reciprocal order, then filter is Linear Phase FIR filter
[E] When zeros of the filter are INSIDE the unit circle filter is called Minimum Phase Filter.
Concept : For Minimum Phase filter φ(π) - φ(0) = 0
[F] When all zeros of the filter are OUTSIDE the unit circle filter is called maximum phase filter.
Concept : For Maximum Phase filter φ(π) - φ(0) = ± m π
30
Kiran Talele ( talelesir@yahoo.com 9987030881 )
[G] When System is Neither Minimum Phase Nor Maximum Phase, Then
System is Mixed Phase System
[H] When all zeros of the FIR filter are LEFT side of POLES, filter is LOW PASS FIR
FILTER.
[I] When all zeros of the FIR filter are RIGHT side of ZEROS, filter is HIGH PASS FIR
FILTER.
[J] When zeros of the FIR filter are Both sides of POLES, Then filter is BAND PASS FIR
FILTER.
[K] When All ZEROS are on Left side of POLES , then filter is LPF
[L] When All ZEROS are on Right side of POLES , then filter is HPF
[M] When ZEROS of the filter are outer sides of POLES, then filter is BPF
(z + 1) (z - 1)
e.g. (i) H ( z ) =
(z - 0.5)(z + 0.5)
31
Kiran Talele ( talelesir@yahoo.com 9987030881 )
[N] When POLES and ZEROS of the filter are in reciprocal order, filter is ALL
PASS FILTER.
Z−2
Eg. IIR filter , H (z) = POLE P1 = 0.5 ZERO Z1 = 2
Z − 0 .5
Here, ZERO = 1/POLE ∴ Filter is All Pass IIR Filter.
The linear Phase characteristic is important when the phase distortion is not tolerable.
FIR Filter can be designed with linear phase characteristic. In application like data
transmission, speech processing etc phase distortion can not be tolerated and here linear phase
characteristic of FIR filter is useful
(176) Show that if the Phase Response is Linear the output of the Filter during pass-band is delayed
input.
Consider a LPF with frequency response H(e–jwα) given by
⎧⎪ e − jwα | w | ≤ wc
H (e jw ) = ⎨
⎪⎩ 0 wc < w ≤ π
x[n] y[n]
H(ejw )
X (w) Y (w)
32
Kiran Talele ( talelesir@yahoo.com 9987030881 )
(177) What is the role of window in the design of FIR filter ? Name the few types of
windows.
Ans : FIR filter is designed by truncating infinite samples of hd[n] by using window
function. Examples of window function include, Hamming window, Bartlet
Window, Hanning window, Blackman window etc.
(178) Why rectangular window is not preferred for FIR filter design ?
Ans : Rectangular window function has As = 21 db which is very small compared to other window
function. Larger value of As desired.
(179) Is the following filter a linear phase filter. If yes, what is the type of filter ? It’s transfer
function is given by H(z) = 1 – z –4 .
Ans : By IZT h[n] = { 1, 0, 0, 0, –1 } Since h[n] is anti-symmetric, filter is a linear phase FIR filter.
Antisymmetric h[n] with N odd is suitable only for Band Pass Filter.
(i) At w = 0, z = 1 : H(w) = 0
(ii) At w = π, z = – 1: H(w) = 0
(iii) At w = π/2, z = j: H(w) = 2
(180) Why antisymmetric h[n] is not suitable for LPF filter design ?
Ans : [ Refer notes ]
(181) Why sy mmetric h[n] with N even and anti-sy mm h[n] with N odd is not
suitable for HPF design ?
Ans : [ Refer notes ]
33
Kiran Talele ( talelesir@yahoo.com 9987030881 )
(186) Why FIR filters are called as Non-recursive filters?
Ans : In FIR filter output depends only on input values. It doesn’t depend on output values.
e.g. y[n] = x[n] + x[n-1]
Therefore FIR filters are also called as Non-Recursive Filters.
(187) Explain how to find output of digital FIR filter in real time application.
Ans : In real time applications, output of FIR filter is obtained using overlap add method / overlap
save method.
(188) Explain how to find output of digital IIR filter in real time application.
Ans : In real time applications, output of IIR filter can be obtained by evaluating difference equation.
(189) Can we use Overlap Add Method and Overlap Save Method to find output of IIR filter for long
data sequence.
Ans : No.
Ans : (i) .The Fourier Transform of the window function W(ejw) should have a small width of
main lobe containing as much of the total energy.
(ii) . The Fourier Transition of the window function W(ejw) should have side lobes that
decrease in energy rapidly as w tends to π .
34
Kiran Talele ( talelesir@yahoo.com 9987030881 )
6. Miscellaneous
Ans : A notch filter is a filter that contain one or more deep notches or ideally perfect nulls in its
frequency response characteristic.
They are useful in application where specific frequency components must be eliminated. For
example instrumentation and recording systems required that the power line frequency of 60
Hz and its harmonics to be eliminated.
Ans : A comb filter can be viewed as a notch filter in which the null occur periodically across the
frequency band.
. Comb filters find applications in a wide range of practical systems such as in the rejection of
power line harmonics, is the separation of solar and lunar components from ionosphere
measurements of electron concentration and is the suppression of cluster from fixed objects in
moving target indicates (MTI) radars.
35
Kiran Talele ( talelesir@yahoo.com 9987030881 )
(200) What is the difference between Pentium Processor and DSP
Processor ? [ Refer Notes ]
As shown in figure, analog input signal is band limited using antialiasing filter which is then
sampled and DT signal thus obtained is converted into digital signal using AdC. Digital processor,
perform the operation depending upon the algorithm programmed in digital processor.
The output of the digital processor is converted inot analog signal using Dac. Reconstruction filter
is used to obtain the corresponding analog signal from the output DT signal.
36
Kiran Talele ( talelesir@yahoo.com 9987030881 )