Unit-V

Multirate Digital Signal

Processing
Multirate Digital Signal Processing
• Introduction, Up-sampler, Down-sampler
• Interpolation and Decimation
• Sampling rate conversion (reduction, increase)
• Sampling rate change by non-integer factor
• Multistage Decimation
• Poly phase structures and implementation
 Reference Books
• Digital Signal Processing, John G. Proakis, and Dimitris G.
Manolakis, 4th Edition, 2007, PHI, ISBN: 81-317-1000-9.
• Digital Signal Processing – Fundamentals and Applications,
Li Tan, 2008, Elsevier, ISBN: 978-0-12-374090-8.
 What is a Multirate Digital Signal
Processing?
• A digital signal processing system that uses signals with
different sampling frequencies is probably performing
multirate digital signal processing.
• A discrete time system with unequal sampling rate at various
parts of the system.
• Multirate DSP often uses sample rate conversion to convert
from one sampling frequency to another sampling frequency.
• Sample rate conversion uses decimation to decrease the
sampling rate, and interpolation to increase the sampling rate.
 Multirate Digital Signal Processing
• Multirate systems use:
– M-fold decimator
– L-fold interpolator
• In most applications, multirate systems are used to:
– Performs the processing in all-digital domain
– Improve the performance (increased computational efficiency)
– Reduce computational complexity
– Reduce transmission rate and/or reduced storage requirements
depending on the application
(Eg. Polyphase implementations)
– Reduce filter order after decimation
 Sample Rate Conversion
• Changing the sampling frequency in the analog domain requires:
– Digital to analog conversion then Analog to digital conversion at
a different sampling frequency.
• Both digital to analog conversion and analog to digital conversion
introduce errors and noise into the signal.
• Sample rate conversion is done in digital domain and uses a
combination of: down-sampler, and up-sampler.
• Down-sampler used to decrease the sampling rate by an integer
factor.
• Up-sampler used to increase the sampling rate by an integer factor.
• The up-sampler and the down-sampler are linear but time-varying
discrete-time systems
 Down-Sampler
Time-Domain Characterization
• A down-sampler with a down-sampling factor M (M is a
positive integer), generates an output sequence y[n] with a
sampling rate that is (1/M)th of that of the input sequence x[n].
• Block-diagram representation is as shown

x[ n ]  xa ( nT ) M y[ n ]  xa ( nMT )

Input sampling frequency Output sampling frequency

1 FT 1
FT  FT 
' 
T M T'
 Down-Sampler
• Down-sampling operation is implemented by keeping every
Mth sample of x[n] and removing (M – 1) in-between samples
to generate y[n].
• Samples equal to multiples of M are retained by down-sampler.
• Input-output relation y[n] = x[nM]
x(n): 8 7 4 8 9 6 4 2 2 5 7 7 6 4 . . . T = 0.1 s

If we downsample the data sequence by a factor of 3, then the

downsampled sequence is
T = 3*0.1 = 0.3 s
y(m): 8 8 4 5 6...,
 Down-Sampler
Frequency-Domain Characterization
• Applying the z-transform to the input-output relation of a
factor-of-M down-sampler
y[n]  x[Mn]
we get 
n
Y ( z)   x[Mn] z
n  
• The expression on the right-hand side cannot be directly
expressed in terms of X(z)
 Down-Sampler
• Aliasing can occur in the downsampled signal due to the
reduced sampling rate.
• After downsampling by a factor of M, the new sampling
period becomes MT, and the new sampling frequency is,

where fs is the original sampling rate.

• The folding frequency after downsampling is,

• The corresponding normalized stop frequency edge is then

converted to be
 Down-Sampler
• Consider a factor-of-2 down-sampler with an input x[n] whose
spectrum is as shown

• The DTFTs of the output and the input sequences of this

down-sampler are then related as
1
Y (e )  { X (e j / 2 ) 
j X (e j / 2 )}
2
 Down-Sampler
• Now X (e j / 2 )  X (e j (2) / 2 ) implying that the second
term X (e j / 2 ) in the previous equation is simply obtained by
shifting the first term X (e j / 2 ) to the right by an amount 2π as
shown below

• Overlap causes the aliasing that takes place due to under-

sampling.
• Hence, the down-sampler causes aliasing.
• Information is lost when a signal is down-sampled.
 Down-Sampler

• The plots of the two terms have an overlap, and hence, in

general, the original “shape” of X (e j ) is lost when x[n] is
down-sampled as indicated below
 Down-Sampler
• Aliasing is absent if and only if
X (e j )  0 for    / M
as shown below for M = 2
X (e j )  0 for    / 2
 Down-Sampler

• For the general case, the relation between the DTFTs

of the output and the input of a factor-of-M down-
sampler is given by,
1 M 1
Y ( e j )   X (e j (2k ) / M )
M k 0
• Y (e j ) is a sum of M uniformly shifted and stretched
versions of X (e j ) and scaled by a factor of 1/M.
 Down-Sampler
• Before downsampling, we can guarantee that the maximum
frequency of the filtered signal satisfies

• Such that no aliasing noise is introduced after downsampling.

• The filtered output in terms of z-transform is W(z) = H(z)X(z)
where X(z) is the z-transform of the sequence to be decimated,
x(n), and H(z) is the lowpass filter transfer function.
• After anti-aliasing filtering, the downsampled signal y(m)
takes its value from the filter output as:
 Decimator
• Prior to down-sampling, the signal should be first bandlimited
to    / M by means of a lowpass filter, called as
decimation filter, as indicated below to avoid aliasing caused
by down-sampling.

x[n] H (z) M y[n ]

• The above system is called a decimator.

 Up-Sampler
• Increasing a sampling rate is a process of upsampling by an
positive integer factor of L, this process is described as
follows:
 Up-Sampler
Time-Domain Characterization
• An up-sampler with an up-sampling factor L, where L is a
positive integer, develops an output sequence xu [n] with a
sampling rate that is L times larger than that of the input
sequence x[n].
• Block-diagram representation
x[ n ]  xa ( nT ) L y[n]
 xa ( nT / L ), n 0,  L, 2 L,

 0 otherwise
Input sampling frequency Output sampling frequency
1 1
FT  FT  LFT 
'
T T'
 Up-Sampler

• Up-sampling operation is implemented by inserting L – 1

equidistant zero-valued samples between two consecutive
samples of x[n].
• Input-output relation
n  0,  L,  2 L,
xu [n]  
x[n / L],
 0, otherwise
 x[n / 2], n  0,  2,  4,
x u [n ]  
 0, otherwise
 Up-Sampler
• In terms of the z-transform, the input-output relation is then
given by
 
n n
X u ( z)   u
x [ n ] z   x[ n / 2] z
n   n  
n even

 
m
x[m] z 2 m  X ( z 2 )

X u ( z )  X ( z ) for a factor-of-L up-sampler

L

X u (e j )  X (e jL )
 Up-Sampler
• Figure below shows the relation between X (e j ) and X u (e j )
for L = 2 in the case of a typical sequence x[n].
 Up-Sampler
• A factor-of-2 sampling rate expansion leads to a compression
of X (e j ) by a factor of 2 and 2-fold repetition in the baseband
[0, 2].
• This process is called imaging as we get an additional “image”
of the input spectrum.
• The up-sampler introduces spectral images.
• Similarly in the case of a factor-of-L sampling rate expansion,
there will be L - 1 additional images of the input spectrum in
the baseband.
• Lowpass filtering of xu [n] removes the L - 1 images and in
effect “fills in” the zero-valued samples in xu [n] with
interpolated sample values.
 Interpolator
• Up-sampling causes periodic repetition of the basic
spectrum, the unwanted images in the spectra of the up-
sampled signal xu [n] must be removed by using a lowpass
filter H(z), called the interpolation filter, as indicated below.
xu [n]
x[n] L H (z) y[n ]

• The above system is called an interpolator.

 Cascade Equivalences

• Two other cascade equivalences are shown below

Cascade equivalence #1
x[n] M H (z ) y1 [ n ]

 x[n] H (z M ) M y1 [ n ]
Cascade equivalence #2

x[n] L H (z L ) y2 [ n]

 x[n] H (z ) L y2 [ n]
Interpolation and Decimation
Multirate Digital Signal Processing
• Introduction, Up-sampler, Down-sampler
• Interpolation and Decimation
• Sampling rate conversion (reduction,
increase)
• Sampling rate change by non-integer factor
• Multistage Decimation
• Poly phase structures and implementation
 Sampling Rate Conversion
• A complex multirate system is formed by an interconnection of
the up-sampler, the down-sampler, and the components of a
digital filter.
• In many applications these devices appear in a cascade form.
• An interchange of the positions of the branches in a cascade
often can lead to a computationally efficient realization.
• To implement a fractional change in the sampling rate we need
to employ a cascade of an up-sampler and a down-sampler.

x[n] M L y1 [ n ]

x[n] L M y2 [ n]
 Filters for Fractional Sampling Rate
Alteration
• A fractional change in the sampling rate can be achieved by
cascading a factor-of-M decimator with a factor-of-L
interpolator, where M and L are positive integers.
• Such a cascade is equivalent to a decimator with a decimation
factor of M/L or an interpolator with an interpolation factor of
L/M.
• There are two possible such cascade connections

H d (z) M L H u (z)

L H u (z) H d (z) M
Multirate Digital Signal Processing
• Introduction, Up-sampler, Down-sampler
• Interpolation and Decimation
• Sampling rate conversion (reduction, increase)
• Sampling rate change by non-integer
factor
• Multistage Decimation
• Poly phase structures and implementation
 Sampling Rate Change by Non-Integer
Factor L/M
• Viewed as two sampling conversion processes:
• Step 1: Perform the upsampling process by a factor of integer
L following application of an interpolation filter H1(z).
• Step 2: Filtering the output from the interpolation filter via an
anti-aliasing filter H2(z), and finally operate downsampling.

• The interpolation and anti-aliasing filters are in a cascaded

form and operate at the same rate, Select one of them.
 Sampling Rate Change by Non-Integer
Factor L/M
• The second scheme is more computationally efficient since
only one of the filters, H u (z ) or H d (z ) , is adequate to serve
as both the interpolation and the decimation filter.
• Hence, the desired configuration for the fractional sampling
rate alteration is as indicated below where the lowpass filter
H(z) has a stopband edge frequency given by,
   
 s  min , 
L M 
L H (z) M
 Multistage Decimation
• The multistage approach for downsampling rate conversion
can be used to dramatically reduce the anti-aliasing filter
length.
• For two-stage decimator, total decimation factor is M = M1 x
M2.
• Develop a procedure for a two-stage case, and a similar
principle can be applied to general multistage cases.
Two-stage Decimator
Two-stage Decimator: Example
 Example
CD Audio to Digital Audio Type System
 CD Audio to Digital Audio Type System

• To convert the CD audio at the sampling rate of 44.1 kHz to

the MP3 or digital audio type system, in which the sampling
rate of 48 kHz is used.
• Conversion factor, L/M = 48000/44100 = 160/147 is required.
• Using the single-stage scheme may cause impractical FIR
filter sizes for interpolation and decimation.
• Since L/M = 160/147 = (8/7)(5/7)(4/3), design an efficient
three stage system
• The stages 1, 2, and 3 use the conversion factors L/M = 8/7,
L/M = 5/7, and L/M = 4/3, respectively.
Sample rate conversion in the CD Audio player system
Multirate Digital Signal Processing
• Introduction, Up-sampler, Down-sampler
• Interpolation and Decimation
• Sampling rate conversion (reduction, increase)
• Sampling rate change by non-integer factor
• Multistage Decimation
• Poly phase structures and
implementation
 Polyphase Filter Structures and
Implementation
• Polyphase filter structures are developed to efficiently
implement the decimation and interpolation filters.
• Polyphase filter structures use fewer numbers of additions and
multiplications.
• The polyphase decomposition of H(z) leads to a parallel form
structure.
• Next we will consider:
– Basic Idea and Examples
– Polyphase filter implementation for the Interpolation
– Polyphase filter implementation for the Decimation
Basic Idea
FIR and IIR Example
 Polyphase Structures/Example
• H(z) can be expressed as a sum of two terms, with one
term containing the even-indexed coefficients and the
other containing the odd-indexed coefficients:

H ( z )  (h[0]  h[2]z 2  h[4]z 4  h[6]z 6  h[8]z 8 )

 (h[1]z 1  h[3]z 3  h[5]z 5  h[7]z 7 )

H(z)  (h[0]  h[2]z 2  h[4]z 4  h[6]z 6  h[8]z 8 )

 z 1(h[1]  h[3]z 2  h[5]z 4  h[7]z 6 )

 Polyphase Structures
• By using the notation
1 2 3 4
E0 ( z )  h[0]  h[2]z  h[4]z  h[6]z  h[8]z
1 2 3
E1( z )  h[1]  h[3]z  h[5]z  h[7]z

• Express H(z) as
1
H ( z )  E0 ( z )  z E1( z )
2 2
 Polyphase Structures
• In a similar manner, by grouping the terms in the
original expression for H(z), re-express it in the form
1 2
H ( z )  E0 ( z )  z E1( z )  z E2 ( z )
3 3 3

where now
1 2
E0 ( z )  h[0]  h[3]z  h[6]z
E1( z )  h[1]  h[4]z 1  h[7]z 2
E2 ( z )  h[2]  h[5]z 1  h[8]z 2
 Polyphase Structures
• The decomposition of H(z) in the form
H ( z )  E0 ( z 2 )  z 1E1( z 2 )
or
H ( z )  E0 ( z 3 )  z 1E1( z 3 )  z 2 E2 ( z 3 )
is known as the polyphase decomposition
 Polyphase Structures/Realization
• Figures show the 4-branch, 3-branch, and 2-branch polyphase
realization of a transfer function H(z).

• The expression for the polyphase components Em (z ) are

different in each case.
• The sub filters are Em ( z L )
 Polyphase Filter Implementation for the
Interpolation
• Consider the interpolation process shown, where L = 2, w(m) is the
upsampled signal and y(m) the interpolated output.

• Assume that FIR interpolation filter has four taps, shown as:

• The filter output is:

• Processing each input sample x(n) requires applying the difference

equation twice to obtain y(0) and y(1), requires eight multiplications
and six additions.
Results of the Direct Interpolation Process
Polyphase Filter Implementation for the
Interpolation
 Polyphase Filter Implementation for the
Interpolation
• Polyphase filter implementation for the interpolation in the
previous Figure (L = 2) is as shown.

• Four multiplications and four additions for processing each

input sample x(n) (computation can be reduced by a factor of L)
• The commutative model for the polyphase interpolation filter is
as shown
Verify output y(1) in the previous Table using the polyphase filter
implementation
Verify output y(1) in the previous Table using the
polyphase filter implementation
 Polyphase Filter Implementation for the
Decimation
• Decimation by a factor of 2 and a three-tap anti-aliasing filter.

• Assuming a three-tap decimation filter, we have

• Obtaining each output y(m) requires processing filter difference

equations twice, resulting in six multiplications and four
additions
Results of Direct Decimation Process
Polyphase Filter Implementation for the
Decimation
 Polyphase Filter Implementation for the
Decimation
• The efficient way to implement a polyphase filter is as shown.

• Polyphase filter implementation for decimation requires three

multiplications and one addition for obtaining each output y(m)
• Commutative model for the polyphase decimator is as shown.
Verify y(1) using the polyphase decimation filter
implementation
Verify y(1) using the polyphase decimation filter
implementation
 Properties of Polyphase Filter
• Each polyphase filter operating at the sampling rate fs (8 kHz) is a
downsampled version of the interpolation filter h(n) operating at
the upsampling rate Lfs (32 kHz, assuming L = 4).
• Consider interpolation FIR filter coefficients h(n), impulse
response sequence having a flat frequency spectrum up to a
bandwidth of fs/2 at the sampling rate of Lfs.
• Downsample h(n) to obtain polyphase filters by a factor of L = 4
and operate them at a sampling rate of fs (8 kHz).
• Nyquist frequency after downsampling is (Lfs/2)/L=fs/2 (4 kHz).
• Each downsampled sequence operating at fs (8 kHz) has a flat
spectrum up to fs/2 (4 kHz) due to the fs/2 (4 kHz) bandlimited
sequence of h(n) at the sampling rate of fs (32 kHz).
• Each polyphase filter has different coefficients, may have a
different phase.
 Summary
• Up-sampler, Down-sampler
• Interpolation and Decimation
• Sampling rate conversion (reduction, increase)
• Sampling rate change by non-integer factor
• Multistage Decimation
• Poly phase structures for interpolation
• Poly phase structures for decimation
THANK YOU