100% found this document useful (1 vote)
4 viewsDsp 5th unit part 1
The document discusses multi-rate digital signal processing, distinguishing between single-rate and multi-rate systems, and detailing the concepts of down sampling and up sampling. It explains the importance of anti-aliasing filters to prevent aliasing during down sampling and the use of anti-imaging filters during up sampling to remove unwanted images. Additionally, it provides examples and mathematical representations of these processes, emphasizing their applications in audio, video, and data acquisition systems.
Uploaded by
csai91957Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
100% found this document useful (1 vote)
4 viewsDsp 5th unit part 1
The document discusses multi-rate digital signal processing, distinguishing between single-rate and multi-rate systems, and detailing the concepts of down sampling and up sampling. It explains the importance of anti-aliasing filters to prevent aliasing during down sampling and the use of anti-imaging filters during up sampling to remove unwanted images. Additionally, it provides examples and mathematical representations of these processes, emphasizing their applications in audio, video, and data acquisition systems.
Uploaded by
csai91957Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
/ 53
Multi-rate Digital Signal Processing
10.1 INTRODUCTION
Discrete-time systems may be single-rate systems or multi-rate systems, The systems that use
single sampling rate from A/D converter to D/A converter are known as single-rate
and the discrete-time systems that process data at more than one sampling rate are known as
multi-rate systems. In digital audio, the different sampling rates used are 32 kHz for
broadcasting, 44.1 kHz for compact dise and 48 kHz for audio tape. In digital video, the
sampling rates for composite video signals are 14.3181818 MHz and 17.734475 MHz for
NTSC and PAL respectively. But the sampling rates for digital component of video signals
are 13.5 MHz and 6,75 MHz for luminance and colour difference signal. Different sampling
rates can be obtained using an up sampler and down sampler. The basic operations in
multirate processing to achieve this are decimation and interpolation. Decimation is for
reducing the sampling rate and interpolation is for increasing the sampling rate. There are
many cases where multirate signal processing is used, Few of them are as follows
In high quality data acquisition and storage systems
In audio signal processing
In video
In speech processing
In transmuttiplexers
6. For narrow hand filtering
The various advantages of multirate signal processing are as follows:
Computational requirements are less.
Storage for filer coefficients is less,
Finite arithmetic effects are less.
Filter onler required in multirate application is low.
Sensitivity to filter coefficient lengths is less.
746While designing multi-rate systems, effects of aliasing for decimation and pscudoimages
for interpolators should be avoided
10.2 SAMPLING
A continuow
it at reg
ime signal x(0) can be converted into a discrete-time signal x(n7) by sampling
tervals of time with sampling period T, The sampled signal x(a7) is given by
x(n) = (0) em cnces
bent”
A sampling proc
so be interpreted as a modulation or multiplication process,
Sampling Theorem
Sampling theorem states that a band limited signal x(t) having finite energy, which has no
spectral components higher than f, hertz can be completely reconstructed from its samples
taken at the rate of 2f, or more samples per second.
The sampling rate of 2/, samples per second is the Nyquist rate and its reciprocal 1/2fj,
is the Nyquist period.
10.3 DOWN SAMPLING
Reducing the sampling rate of a discrete-time signal is called down sampling. The sampling
rate of the discrete-time signal can be reduced by a factor D by taking every Dth value of
the signal, Mathematically. down sampling is represented by
y(n) = (Dn)
and the symbol for the down sampler is shown in Figure 10.1
igure 10.1 A down sampler.
If x(n) = (1,2, 3, 1,2, 3, 1, 2, 3, a)
‘Then, x(2n) 13. ad
and xGn) Tyra)
x(2n) is obtained by keeping every second sample of x(n) and x(3n) is obtained by keeping
every 3rd sample of (7) and removing other samples.
If the input signal x07) is not band limited, then there will be overlapping of spectra at
the output of the down sampler. This overlapping of spectra is called aliasing which is
undesirable. This aliasing problem can be eliminated by band limiting the input signal by
inserting a low-pass filter called anti-aliasing filter before the down sampler. The anti-
aliasing filter and the down sampler together is called decimator. The decimator is alsoKnown as sub sampler, down sampler or under sampler. Decimation (sampling rate
compression) is the process of decreasing the sampling rate by an integer factor D by
keeping every Dth sample and removing D—1 in between samples
Figure 10.2 shows the signal x(n) and its down sampled versions by a factor of 2 and 3.
fitstitds
oe
Figure 10.2 Plots of (a) xin), (b) x!2a) and (c) x@3n
The block diagram of the decimator is shown in Figure 10.3. The decimator comprises
jasing filer and down sampler. Here the anti-aliasing filter is a
aliasing problem is eliminated and the
fe by keeping every Dth sample and
two blocks such as anti-
low-pass filter to band limit the input signal so th;
down sampler is used to reduce the sampling 1
removing D—1 in between samples.
Inpat urput
signal maton ]_vn signal
oo a [Down sampler me
iter hin) or Ho) iD
Figure 10.3. Block diagram of decimator.
Spectrum of down sampled signal
Let T be sampling period of input signal x0), and let F be its sampling rate or frequency
When the signal is down sampled by D, let 7” be its new sampling period and F” be its
sampling frequency, then
rep
T
Lite
pel Lt
TD" D
Let us derive the spectrum of a down sampled signal x(Dn) and compare it with the
spectrum of input signal a(n). The Z-transform of the signal x(n) is given byX@= DY xumer
The down sampled signal yQ1) is obtained by multiplying the sequence x(n) with a periodic
train of impulses p(n) with a period D and then |
pair of samples. The periodic train of impulses is given by
ing out the D1 zeros between each
1, n=0, £D, 2D,
pan)=
0, otherwise
The discrete Fourier series representation of the signal p(t) is given by
1 a
poy = eA, co cn cos
im
Multiplying the sequence a(n) with pin) yields
vn) = xinyp(n)
[xu =0, £D, £2...
That is v=
JO, otherwise
If we leave D1 zeros between cach pair of samples, we get the output of down
sampler
Nn) = x(n) = x(n) pinDy
= xnD)
The Z-transform of the output sequence is given by
y= Y yoye"
5 vomer
Loe”
where x(n) = 0 except at multiple of D. Since x(n) = x(n) pin), we get
Y= Y xy poem”(ny (PRMD ID
5 ein
D
Substituting z = e!, we get the frequency response
LS yee stmt gsany RS yeti
Yana & X¢ ) pe y
in ont)
ie, Y@=— ¥ xi-—
From the above relation we find that if the Fourier transform of the input signal x(n) of a
down sampler is X(@), then the Fourier transform ¥(@) of the output signal y(n) is a sum of
D uniformly shifted and stretched versions of X(@) scaled by a factor 1/D.
If the spectrum of the original signal X(@) is band limited to @ = 72/d, as shown in
Figure 10.4(a), the spectrum being periodic with period 27, the spectrum of the down
sampled signal ¥(@) is the sum of all the uniformly shifted and stretched versions of X(@)
scaled by a factor I/D as shown in Figure 10.4(b). In every interval of 27 in addition to the
original spectrum we find D1 equally spaced replica. In Figure 10.4(b), the frequency
Variable @, is related to the original sampling rate, In Figure 10.4(¢), the frequency variable
©, is normalized with respect to reduced sampling rate,
eo
ot a =,
(Mao)
aD wb wi ay
®)
i i or De
©
Figure 10.4 Spectrum of (a) input, (b) output, and (c) normalized! outputAliasing effect and Anti-aliasing filter
From Figure 10.5, we ean find that the spectrum obtained after down sampling will overlap
if the original spectrum is not band limited to @ = 2/D. This overlapping of spectra is called
aliasing, Therefore, aliasing due to down sampling a signal by a factor of D is absent if and
only if the signal x(n) is band limited to 422/D. If the signal x(n) is not band fimited to
27/D, then a low-pass filter with a cutoff frequency 2/D is used prior to down sampling.
This low-pass filter which is connected before the down sampler to prevent the effect of
aliasing by band limiting the input signal is called the anti-aliasing filter.
ial
0 & oF wo,
DD
mat
on a Aliasin, 2% oh
DD s
Figure 10.5 (a) Input spectrum, (b) aliased output spectrum,
The signal obtained after filtering is given by
van
> bb an-k)
‘
and
Y dyed)
For example, consider a factor of D down sampler, then‘The second term X(-/2) is simply obtained by shifting the first term X(@) to the right
by an amount of 2.
104 UP SAMPLING
Increasing the sampling rate of a discrete-time signal is called up sampling. The sampling
rate of a discrete-time signal can be increased by a factor I by placing 1 ~ 1 equally spaced
zeros between each pair of samples,
Mathematically, up sampling is represented by
af), nso, 44,
vn) = (7)
0, otherwise
and the symbol for up sampler is shown in Figure 10.6.
xy) vin)
7 Pe
Figure 10,6 Up sampler.
u xv) = (1.2.3, 1 2, 3 ad
Then, yin) = (3) (1.0, 2.0.5.0, 1,0 2.0, 3.0, 4.) for an up-sampling factor of J
and y(n (5) (1, 0,0, 2,0, 0, 3,0,0, 1, vu) for an up-sampling factor of 1
Usually an ancisimaging filter la to be: kept after the up sampler to remove the
unwanted images developed dive to up sampling, The ant-imaging filter and the up sampler
together is called interpolator. Interpolation is the process of Inereasing the sampling rate by
integer factor by interpolating #1 new samples between succesive values of the signal
Figure 10.7 shows the signal x(n) and its two-fold up-sampled signal yy(m) and the
interpolated signal y(n).
The block diagram of the interpolator is shown in Figure 10.8, The interpolator
comprises two blocks such as up sampler and anti-imaging filter, Here up sampler is used to
increase the sampling rule by inirtucing zeros between succeasive ingut samples and the
interpolation filter, also known as ant-imaging fle, is used to remove the unwanted images
that are yielded by up sampling.
Expression for output of interpolator
Let J be an integer interpolating factor of the signal. Let T be sampling period and F = VT
be the sampling frequency (sampling rate) of the input signal, After up sampling, let T” be
the new sampling period and F” be the new sampling frequency, thenxm of 1
oa
u
z
Figure 10.7 (a) Input signal xin, ¢b| Output of 2 fold up sampler y(n) = x/2), (¢ Output of interpobtor
vyatn) = xini2),
np Up sampled Out
al Trapaen —|_ sil
ain) ae Up sampler finery [—F, 3
Figure 10.8 Block diagram of an interpolator.
‘The sampling rate is given by
Let s(n) be the signal obtained by interpolating 1 samples between each pair of samples
of x.
‘The Ztransform of the signal w(x) is given by
w= 5 wine784 2 Digital Signal Processing
= Dame
X(2!)
lered over the unit circle z =e”,
Wee") =X(eP") ic. WO") = XU)
where (= 2/7". The spectra of the signal w(1) contains the images of base band placed at
the harmonics of the simpling frequency +27, 44M/1. To remove the images an anti-
imaging filter is used. The ideal characteristics of low-pass filter is given by
He”) G, |o|> 1 and/or I >> 1, we go in
for multistage implementation
Consider a system for decimating a signal by an integer factor D. Let the input signal
sampling frequency be f,, then the decimated signal frequency will be f, = f/D. If D >> 1
then we express D as a product of positive integers as
Each decimator D, is implemented and cascaded to get V stages of filtering and decimators.
as shown in Figure 10.42,TIR_ 22 Digital Signal Provessing
OF To 7b
eh he hele
D=D,...0,=[]>,
Figure 10.42 Multistage decimator
Similarly, if the interpolation factor 7 >> 1, then express J as a product of positive
integers as:
ided to get N stages of implementation and
Then each interpolator /, is implemented and east
filtering as shown in Figure 10.43,
ee ee
to th fo
igure 10.43 Multistage interpolation,
n, the computational
If the sampling rate alteration system is designed as a cascade syst
efficiency is improved significantly. The reasons for using multistage structures are as Follows:
L. Multistage system requires less computation,
2. Storage space required is less.
3. Filter design problem is simple.
4. Finite wonl length effects are less
ystems are that proper control structure is required in implen
The demerits of the s
ting the system and proper values of should be chosen.Mulirate Digital Signal Processing = 779.
EXAMPLE 10.12 For the multi-rate system shown in Figure 10.44, develop an expression
for the output y(7) as a function of the input x(n),
xo) 14 12 fis vo)
Figure 10.44 Example 10.12.
Solution: In Figure 10.45 the down sampler with D = 12 is split into 2 down samplers with
D, = 3 and Dy = 4. The up sampler with 7 = 4 and the down sampler with D = 3 are
interchanged and finally the up sampler with f = 4 and down sampler with D = 4 cancel, and
wwe will be left with a down sampler with D = 3 and an up sampler with J =
x(n 4 2 13 on)
=x0) 14 43 44 13 v0)
2x60 43 14 14 13 ven)
SE ee
FIGURE 10.45. Simplification of Figuie 10.44.
From Figure 10.45, we have
ain) = Gn)
5} «:
otherwise
and ya)
s(n), for n= 3k
ie wn= \
0, otherwise
EXAMPLE 10.13 A multirate syst
x(n) and yn.
is shown in Figure 10.46, Find the relation between
Solution: The given system is shown in Figure 10.47.
In Figure 10.47 after down sampling x(n), we get
vn) = x2n)780 Digital Signal Processing
42 12
Figure 10.46 Example 10.13.
ee ee
Figure 10.47 Example 10.13.
If we up sample v(n) by 2, we get
&
vl i
0, otherwise
‘) fo for n=0. £2,
If we delay x(1) and down sample, we get
wa) = an 1)
If we up sample w(n), we get
[xn=D, for n=
ial Jo, otherwise
yn) = y(n 1) + wyla)
gle) = {2(0), 0,2(2),0,403), 00}
¥,(n=1) = (0, x(0), 0, 2(2),0, (3), -.}
wr) = [x(-1),0, x(1),0, 963), 0.
wd) + vy (21) = (x= 1), 1), x1), 202), A)
yin) = x= 1)Mutat Digital Signal Processing 781
10.10 EFFICIENT TRANSVERSAL STRUCTURE FOR DECIMATOR
We discussed earlier that a decimator consists of an anti-aliasing filter followed by a down
sampler as shown in Figure 10.48.
we
xin) Pn) as vin)
Figure 10.48 A decimator:
Lot us assume that the anti-aliasing filter is an FIR filter with N’ coefficients, The
‘output v(n) of an FIR filter is the convolution of input x(») and impulse response hr) and is
given by
vn) = a(n) * hn)
Yiyx@-6
B
The output v(7) is then down sampled to yield
yn) = (nD)
= Yitoxnd—h)
7
FIR filters are nom
by
ily realized with linear phase. Hence impulse response is symmet
iven
Wk) = WN — 1b
With this property, the number of multiplications can he reduced by a factor of two. If V is
'
= Shon) +N
fy
von
a)
‘The FIR filter can be realized using direct form structure as shown in Figure 10.49. TI
direct form realization, shown in Figure 10.49, is very inefficient as it involves the
calculation of even the interim values of y(n) which are not used later on.
To avoid unnecessary calculation of the values of v(m), m # nD an efficient transversal
structure shown in Figure 10.50 is used. Here the multiplications and additions a
performed
at reduced sampling rate782_ 22 Digital Signal Processing
at 0) nm)
7 0) nd
Figure 10.49 Direct form wulization of a decimator,
x0) 00) vin)
may
io
ToL hm
'
t Lo
pL mee
Figure 10.50. Etfcient realization for decimator.
10.11 EFFICIENT TRANSVERSAL STRUCTURE FOR INTERPOLATOR
Earlier we discussed that an interpolator consist
imaging filter h(n) (a low-pass filter) as shown in
of an up sampler f followed by anti-
re 10.51
xo) yy LD xen)
Figure 10.51 An interpolator,Muthinate Digital Stenat Processing 22783
The transposed dire
up sampling produces
ith the impulse response /n(n). If the ant
ficients, then output
form structure using an FIR filter is shown in Figure 10.52. The
interim signal v(2), The output signal y(7) is obtained by convolving
naging filter is an FIR low-pass filter with
wet
Sven bik)
von)
xn)
10) wn)
Figure 10.52 Transposed dizect form realization for interpolator,
In the process of obtaining y(a) for different values of n, unnecessary calcul
carried out due to zeros inserted because of up ssmpling. So an efficient transver
which avoids these unnecessary computations, is shown in Figure 10.53.
Fa ho) xn)
med
Figure 10.53. Eiicient alization of interpolator.784 = Digital Signal Processing
Duality
From Figures 10.50 and 10.53, we can observe that the structure for interpolator ean be
‘obtained by transposing the structure of the decimator, That is the transpose of a decimator is,
an interpolator, and vice versa This duality relationship between an interpolator and a
decimator is shown in Figure 10.54
i) pur Oupar sie) —_Ovtpu Inga
aja et to —
Input Output orp Input
Ww
us
Figure 10.54 Duality between interpolator and decimator.
10.12 HR STRUCTURES FOR DECIMATORS:
The UR filter is represented by the differen
=D aya-+ SD bxtn-’)
im io
Applying Z-tansform,
u ns
¥Q=Y getvo+ Y exo
fs io
icc. YOU - Yaz=¥ ketx
rt iso
So the system function for the above difference equation is given by
Let M jor and denominator orders are same, Figure 10.55
shows the dir
form of the TIR structure for a D to 1 decimator.Mulirate Digital Signal Processing 785
xn) in ii a
e on D
F AD
UR fier
x),
nt Py
Figure 10.55. The direct form of the IR structure for aD to 1 decimator.
10.13 FILTER DESIGN FOR FIR DECIMATORS AND INTERPOLATORS
The FIR filter design has already been discussed in Chapter 9. There are various methods tor
FIR filter design like window method, optimal equiripple linear phase method, half band
design, etc. Let us consider the equiripple FIR filter design.
The design equations for calculating the stop band and pass band frequency are
discussed below.
Let the highest frequency of the decimated si
interpolated signal be @, $ 7 then the pass band freq
nal or the total bandwidth of the
y is given by
fost, Lo Fimepotar
‘The stop band frequency is given by
fe 1 to 7 interpolator
mI, Dio decimator
min (8/1_22/D), conversion by iD
‘The assumption is that there is no aliasing in the decimator or imaging in the interpolator. If
aliasing is allowed in the deeimator or interpolator, then the stop band frequency is given by786_= Digital Signal Processing
Qn-0,/l, to interpolator
©, = }2n-@,)D. D to 1 decimator
min|[2 ~ @,)/1, 2% ~ @,)/D)], conversion by 1D
10.14 FILTER DESIGN FOR UR INTERPOLATORS AND DECIMATORS
‘The IR filter design has already been discussed in Chapter 8. The ideal characteristic for the
IIR prototype filter hn) for a D to 1 decimator, assuming that no constraints are set for the
phase is given by
eP@, lols
HiBhe |e" @, |olsao
Jo, otherwise
For an interpolator the ideal characteristic becomes,
reo), lost
0. otherwise
H(@)=
The system function for an IIR filter is given by
Represent the decimator polynomial as a polynomial of order R in 2°, where D is the
decimation factor, Replace
mB at
For IIR filter designs, the following approximations are used:
1. The Butterworth approximation
2. The Bessel approximation
3. The Chebyshev approximation
4. The Elliptic approximation
EXAMPLE 10.14 Design one-stage and two-stage interpolators to meet the following
specifications, 7 = 20,
(a) Pass band os Fs9
(b) Transition band 90
