Professional Documents
Culture Documents
3015DigitalSignalProcessingand
Filtering
8.4.2013
Introduction
Finite wordlength effects are caused by:
T-61.3015
Digital Signal Processing and Filtering
Chapter 12:
Analysis of Finite Wordlength Effects
Sanjit K. Mitra,
Digital Signal Processing, A Computer-Based Approach,
3rd Edition, McGraw-Hill, 2006
Olli Simula
8.4.2013
T-61.3015 Digital Signal Processing Filtering; Chapter 12
Chapter12:AnalysisofFiniteWordlength
Efeects
T61.3015DigitalSignalProcessingand
Filtering
8.4.2013
Quantizer:
The quantization
process model
Quantization of coefficients :
Q(x)
Rounding /
Truncation:
Twos complement
truncation
Quantization Error
Rounding
To be discarded
Chapter12:AnalysisofFiniteWordlength
Efeects
10
11
12
T61.3015DigitalSignalProcessingand
Filtering
8.4.2013
Passband Details
14
15
Passband Details
16
Passband details
Gain, dB
0
-20
-40
-60
0.2
0.4
0.6
0.8
Amplitude responses
17
Pole-zero locations
18
Chapter12:AnalysisofFiniteWordlength
Efeects
T61.3015DigitalSignalProcessingand
Filtering
8.4.2013
Example:
Amplitude responses
Pole-zero locations
19
20
Analog input
H(z)
+
E(z)
21
Analog input
sample x[n]
Binary equivalent
Quantized
input sample of quantized input
22
Quantization Error
Twos complement
representation
Chapter12:AnalysisofFiniteWordlength
Efeects
23
24
T61.3015DigitalSignalProcessingand
Filtering
8.4.2013
(a) Rounding
Assumptions:
1) The error sequence {e[n]} is a sample sequence of a widesense stationary (WSS) white noise process, with each sample
e[n] being uniformly distributed over the quantization error
2) The error sequence is uncorrelated with its corresponding input
sequence {x[n]}
3) The input sequence is a sample sequence of a stationary
random process
25
where
x2 is the signal variance (power) and
e2 is the noise variance (power)
27
26
A/D conversion:
(b+1) bits: = 2-(b+1)RFS , where RFS is the fullscale range
28
Chapter12:AnalysisofFiniteWordlength
Efeects
Signal-to-Noise Ratio
Signal-to-Noise Ratio
(b) Twos
complement
truncation
29
30
T61.3015DigitalSignalProcessingand
Filtering
8.4.2013
31
32
Analysis of
Assumptions:
33
Quantization before
summation
The number of
multiplications kl at adder
inputs
The rth branch node with
signal value ur[n] needs to
be scaled to prevent
overflow
T-61.3015 Digital Signal Processing Filtering; Chapter 12
Chapter12:AnalysisofFiniteWordlength
Efeects
35
34
fr[n]
Impulse response
from filter input to
branch node r
gl[n]
Impulse response
from input of lth adder
to filter output
36
T61.3015DigitalSignalProcessingand
Filtering
8.4.2013
37
38
39
40
Digital filter
Chapter12:AnalysisofFiniteWordlength
Efeects
An Absolute Bound
Digital filter
41
42
T61.3015DigitalSignalProcessingand
Filtering
8.4.2013
An Absolute Bound
Assuming that x[n] satisfies the dynamic range
constraint |x[n]| < 1
43
Scaling Norms
Define the Lp-norm of a Fourier transform F(ej) as
44
45
Chapter12:AnalysisofFiniteWordlength
Efeects
46
47
or equivalently
If the filter input has finite energy bounded by unity,
i.e., ||X||2 < 1, then the adder overflow can be
prevented by scaling the filter such that the RMS
value of the scaling transfer functions are bounded
by unity:
T-61.3015 Digital Signal Processing Filtering; Chapter 12
48
T61.3015DigitalSignalProcessingand
Filtering
8.4.2013
Scaling - Back-Scaling
50
FILTER
Scaling transfer
functions:
The effect of input scaling is compensated by backscaling at the output of the filter
Scaling block-by-block in cascade realization forms
H1(z)
H2(z)
HR(z)
51
52
Chapter12:AnalysisofFiniteWordlength
Efeects
53
54
T61.3015DigitalSignalProcessingand
Filtering
8.4.2013
kl = 1, for l = 1, 2,...,R+1
56
57
58
Section Ordering
Chapter12:AnalysisofFiniteWordlength
Efeects
55
59
=>
60
10
T61.3015DigitalSignalProcessingand
Filtering
8.4.2013
Section Ordering
L2 scaling:
The ordering of paired sections does not influence too much
the output noise power since all norms in the expressions
are L2-norms
L scaling:
The sections with poles closest to the unit circle exhibit a
peaking magnitude response and should be placed closer to
the output end
=> Ordering should be from least-peaked to most-peaked
On the other hand, the ordering scheme is exactly opposite if
the objective is to minimize the peak noise ||Pyy()|| and L2scaling is used
The ordering has no effect on the peak noise with L-scaling
T-61.3015 Digital Signal Processing Filtering; Chapter 12
61
62
63
Substituting w[n]:
64
Chapter12:AnalysisofFiniteWordlength
Efeects
65
66
11