Discrete-time Processing of Continuous-time Signals
Transcript of Discrete-time Processing of Continuous-time Signals
![Page 1: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/1.jpg)
Discrete-time Processing of Continuous-time Signals
(cf. Oppenheim, 1999)
A major application of discrete-time systems is in the processing of continuous-time signals.
The overall system is equivalent to a continuous-time system, since it transforms the continuous-time input signal xs(t) into the continuous time signal yr(t).
Question: what is this equivalent system?
![Page 2: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/2.jpg)
Ideal discrete-to-continuous (D/C) converter
![Page 3: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/3.jpg)
Ideal reconstruction filter
)//()/sin()( TtTtthr ππ=
The ideal reconstruction filter is a continuous-time filter, with the frequency response being Hr(jΩ) and impulse response hr(t).
From now on, we use Ω to represent the transform domain of continuous Fourier transform.
![Page 4: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/4.jpg)
Continuous to discrete (C/D) converter
( ) ( )∑∞
−∞=
−==k
cs nTttxtstxtx δ)()()(
( ) ( )∑∞
−∞=
−=k
nTtts δ ( ) ( )∑∞
−∞=
Ω−Ω=Ωk
skT
jS δπ2continuous F. T.
where Ωs = 2π/TTime domain multiplication
![Page 5: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/5.jpg)
C/D converter
We have
Hence, the continuous Fourier transforms of xs(t) consists of periodically repeated copies of the Fourier transform of xc(t).
Review of Nyquist sampling theorem:Aliasing effect: If Ωs< 2ΩN, the copies of Xc(jΩ) overlap, where ΩN
is the highest nonzero frequency component of Xc(jΩ). ΩN is referred to as the Nyquist frequency.
( ) ( )( )∑∞
−∞=
Ω−Ω=Ωk
scs kjXT
jX 1where Ωs = 2π/T
)()(21)( Ω∗Ω=Ω jSjXjX cs π
Frequency domain convolution
![Page 6: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/6.jpg)
In the above, we characterize the relationship of xs(t) and xc(t) in the continuous F.T. domain.From another point of view, Xs(jΩ) can be represented as the linear combination of a serious of complex exponentials:
If x(nT) ≡ x[n], its DTFT is
Ideal C/D converter
( ) ( ) ( )∑∞
−∞=
−=n
cs nTtnTxtx δ since
( ) ( )∑∞
−∞=
Ω−=Ωn
Tnjcs enTxjX
∑∑∞
−∞=
−∞
−∞=
− ==n
jwn
n
jwnjw enTxenxeX )(][)(
![Page 7: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/7.jpg)
Combining these properties, we have the relationship between the continuous F.T. and DTFT of the sampled signal:
where : represent continuous F.T.: represent DTFT
Thus, we have the input-output relationship of C/D converter
Input-output relationship of C/D converter
( ) ∑∑∞
−∞=
∞
−∞=⎟⎟⎠
⎞⎜⎜⎝
⎛⎟⎠⎞
⎜⎝⎛ −=⎟⎟
⎠
⎞⎜⎜⎝
⎛⎟⎠⎞
⎜⎝⎛ −Ω=
kc
kc
jw
Tk
TwjX
TTkjX
TeX ππ 2121
( )jweX( )ΩjX c
( ) ( ) ( )TjTw
jws eXeXjX Ω
Ω===Ω
![Page 8: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/8.jpg)
Consider again the discrete-time processing of continuous signals
Let H(ejw) be the frequency response of the discrete-time system in the above diagram. Since Y(ejw) = H(ejw)X(ejw)
Combine C/D, discrete-time system, and D/C
( ) ( ) ( )TjTjTj eXeHeY ΩΩΩ =
![Page 9: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/9.jpg)
An ideal low-pass filter Hr(jΩ) that has a cut-off frequency Ωc= Ωs/2 = π/T and gain T is used for reconstructing the continuous signal.Frequency domain of D/C converter: (Hr(jΩ) is its frequency response)
Remember that the corresponding impulse response is a sinc function, and the reconstructed signal is
D/C converter revisited
( ) [ ] ( )( )( )∑
∞
−∞= −−
=n
r TnTtTnTtnyty
//sin
ππ
( ) ( ) ( ) ( )⎩⎨⎧ <Ω
=Ω=ΩΩ
Ω
otherwiseTeTY
eYjHjYTj
Tjrr ,0
/, π
![Page 10: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/10.jpg)
Input-output Relationship for Discrete-time processing of continuous-time signals
Assumption:If Xc(jΩ) is band limited with Xc(jΩ) = 0 for |Ω|>π/T, then
and hence
( ) ( ) ( ) ( ) ( )
( ) ( )⎩⎨⎧ <ΩΩ
=
⎩⎨⎧ <Ω
=Ω=Ω
Ω
ΩΩΩ
otherwiseTjXeH
otherwiseTeXeTH
eYjHjY
cTj
TjTjTj
rr
,0/,
,0/,
π
π
( ) ( )Tjc eX
TjX Ω=Ω
1
![Page 11: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/11.jpg)
Effective Frequency Response for Discrete-time processing of
continuous-time signals
So, if Xc(jΩ) is band limited with Xc(jΩ) = 0 for |Ω|>π/T, we have the following effective response for the entire system:
( ) ( ) ( )
( ) ( )⎪⎩
⎪⎨⎧
≥Ω<Ω=Ω
ΩΩ=Ω
Ω
TTeHjH
jXjHjYTj
eff
ceffr
/,/,
,
ππ
0 where
![Page 12: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/12.jpg)
Effective Frequency Response
It is important to emphasize that the LTI behavior of the system depends on two factors:
First, the discrete-time system is LTI.Second, the input signal is band-limited, and the
sampling rate is high enough so that any aliased components are removed.
![Page 13: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/13.jpg)
Discrete-time processing of continuous-time signals
If we are given a desired continuous-time system with band-limited frequency response Hc(jΩ), and we want to implement it by discrete-time processing,
We can choose appropriate T and discrete-time filter satisfying H(ejw) = Hc(jw/T) to synthesize the continuous response Hc(jΩ).
![Page 14: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/14.jpg)
Time-domain behavior of discrete-time processing of continuous-time signals
In time domainSince H(ejw) = Hc(jw/T) |w| < πIn addition, Hc(jΩ) = 0, |Ω|>π/T (band limited)We have the impulse invariance property, h[n]=Thc(nT),i.e., the impulse response of the discrete-time system is a scaled, sampled version of the continuous inpulseresponse hc(t).
Remark: It is because that if
and thus since band limited.
)(][ nTThnh c=
π<= || )()( wTwjHeH c
jw
∑∞
−∞=
−=k
cjw
Tk
TwjHeH ))2(()( π
![Page 15: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/15.jpg)
Continuous-time processing of discrete-time signals
On the other hand, we can also consider to process discrete-time signal with continuous-time filters.Cascading D/C, continuous-time system, and C/D.From the definition of the ideal D/C converter, Xc(jΩ) and therefore also Yc(jΩ), will necessarily be zero for |Ω|>π/T.
![Page 16: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/16.jpg)
Continuous-time processing of discrete-time signals
( ) π<⎟⎠⎞
⎜⎝⎛= || 1 w
TwjY
TeY c
jw
( ) ( ) ( ) TjXjHjY ccc /|| π<ΩΩΩ=Ω
( ) ( ) TeTXjX Tjc /|| π<Ω=Ω Ω
( ) π<⎟⎠
⎞⎜⎝
⎛= wTwjHeH c
jweff ,
Hence, we have the effective system of the continuous-time processing of discrete-time signals to be
![Page 17: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/17.jpg)
Changing the Sampling rate using discrete-time processing
downsampling; sampling rate compressor;
[ ]nMxnxd =][
![Page 18: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/18.jpg)
Frequency domain of downsampling
Since this is a ‘re-sampling’ process. Remember that, from continuous-time sampling of x[n]=xc(nT), we have
Similarly, for the down-sampled signal xd[m]=xc(mT’), (where T’ = MT), we have
( ) ∑∞
−∞=⎟⎠⎞
⎜⎝⎛ −=
kc
jw
Tk
TwjX
TeX )2(1 π
( ) ∑∞
−∞=⎟⎠⎞
⎜⎝⎛ −=
rc
jwd T
rTwjX
TeX )
'2
'(
'1 π
![Page 19: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/19.jpg)
Frequency domain of downsampling
We are interested in the relation between X(ejw) and Xd(ejw). Let’s represent r as r = i + kM, where 0 ≤ i ≤M−1, (i.e., r ≡ i (mod M)). Then
( )
∑ ∑
∑−
=
∞
−∞=
∞
−∞=
⎟⎠⎞
⎜⎝⎛ −−=
⎟⎠⎞
⎜⎝⎛ −=
1
0)22(11
)2(1
M
i kc
rc
jwd
MTi
Tk
MTwjX
TM
MTr
MTwjX
MTeX
ππ
π
)()22(1 /)2( Miwj
kc eX
Tk
MTiwjX
Tπππ −
∞
−∞=
=⎟⎠⎞
⎜⎝⎛ −
−= ∑
![Page 20: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/20.jpg)
Frequency domain of downsampling
Therefore, the downsampling can be treated as a ‘re-sampling’ process. It s frequency domain relationship is similar to that of the D/C converter as:
This is equivalent to compositing M copies of the of X(ejw), frequency scaled by M and shifted by inter multiples of 2π.The aliasing can be avoided by ensuring that X(ejw) is bandlimited as
( ) ( )( )∑−
=
−=1
0
21 M
i
Miwjjwd eX
MeX /π
( ) NNjw wwweX 2/M2 and ,||for 0 ≥≤≤= ππ
![Page 21: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/21.jpg)
Example of downsampling in the Frequency domain (without aliasing)
Sampling with a sufficiently large rate which avoids aliasing
![Page 22: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/22.jpg)
Example of downsampling in the Frequency domain (without aliasing)
Downsampling by 2 (M=2)
![Page 23: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/23.jpg)
Downsampling with prefiltering to avoid aliasing (decimation)
From the above, the DTFT of the down-sampled signal is the superposition of M shifted/scaled versions of the DTFT of the original signal.To avoid aliasing, we need wN<π/M, where wN is the highest frequency of the discrete-time signal x[n].
Hence, downsampling is usually accompanied with a pre-low-pass filtering, and a low-pass filter followed by down-sampling is usually called a decimator, and termed the process as decimation.
![Page 24: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/24.jpg)
Up-sampling
Upsampling; sampling rate expander.
or equivalently,
In frequency domain:
[ ] [ ]⎩⎨⎧ ±±=
=otherwise0
2 0,
...,,,,/ LLnLnxnxe:
[ ] [ ] [ ]∑∞
−∞=
−=k
e kLnkxnx δ
( ) ( )jwL
k
jwLk
n k
jwnjwe eXekxekLnkxeX ∑∑ ∑
∞
−∞=
−∞
−∞=
∞
−∞=
− ==−= )][(])[][( δ
![Page 25: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/25.jpg)
Example of up-sampling
Upsampling in the frequency domain
![Page 26: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/26.jpg)
Up-sampling with post low-pass filtering
Similar to the case of D/C converter, upsampoling is usually companied with a post low-pass filter with cutoff frequency π/L and gain L, to reconstruct the sequence.A low-pass filter followed by up-sampling is called an interpolator, and the whole process is called interpolation.
![Page 27: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/27.jpg)
Example of up-sampling followed by low-pass filtering
Applying low-pass filtering
![Page 28: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/28.jpg)
Similar to the ideal D/C converter,If we choose an ideal lowpass filter with cutoff frequency π/L and gain L, its impulse response is
Hence
Interpolation
[ ] ( )Ln
Lnnhi //sin
ππ
=
[ ] [ ] [ ] [ ] [ ] [ ]
[ ] ( )[ ]( )∑
∑∞
−∞=
∞
−∞=
−−
=
∗⎟⎟⎠
⎞⎜⎜⎝
⎛−=∗=
k
ik
iei
LkLnLkLnkx
nhkLnkxnhnxnx
//sin
ππ
δ
Its an interpolation of the discrete sequence x[k]
![Page 29: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/29.jpg)
Sampling rate conversion by a non-integer rational factor
By combining the decimation and interpolation, we can change the sampling rate of a sequence.
Changing the sampling rate by a non-integer factor T’ = TM/L.Eg., L=100 and M=101, then T’ = 1.01T.
![Page 30: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/30.jpg)
Changing the Sampling rate using discrete-time processing
Since the interpolation and decimation filters are in cascade, they can be combined as shown above.
![Page 31: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/31.jpg)
Digital Processing of Analog Signals
Pre-filtering to avoid aliasingIt is generally desirable to minimize the sampling rate.Eg., in processing speech signals, where often only the low-frequency band up to about 3-4k Hz is required, even though the speech signal may have significant frequency content in the 4k to 20k Hz range.Also, even if the signal is naturally bandlimited, wideband additive noise may fll in the higher frequency range, and as a result of sampling. These noise components would be aliased into the low frequency band.
![Page 32: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/32.jpg)
Over-sampled A/D conversion
The anti-aliasing filter is an analog filter. However, in applications involving powerful, but inexpensive, digital processors, these continuous-time filters may account for a major part of the cost of a system.Instead, we first apply a very simple anti-aliasing filter that has a gradual cutoff (instead of a sharp cutoff) with significant attenuation at MΩN. Next, implement the C/D conversion at the sampling rate higher than 2MΩN. After that, sampling rate reduction by a factor of M that includes sharp anti-aliasing filtering is implemented in the discrete-time domain.
![Page 33: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/33.jpg)
Using over-sampled A/D conversion to simplify a continuous-time anti-aliasing filter
![Page 34: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/34.jpg)
Example of over-sampled A/D conversion
![Page 35: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/35.jpg)
Example of over-sampled A/D conversion
![Page 36: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/36.jpg)
Sample and hold
∑∞
−∞=
−=n
nTthnxtx )(][)( 00⎩⎨⎧ <<
=otherwise
Ttth
001
)(0
![Page 37: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/37.jpg)
Example of sample and hold
![Page 38: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/38.jpg)
Quantizer (Quantization)
The real-valued signal has to be stored as a code for digital processing. This step is called quantization.
The quantizer is a nonlinear system.Typically, we apply two’s complement code for representation.
])[(][ˆ nxQnx =
![Page 39: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/39.jpg)
Quantizer (Quantization)
![Page 40: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/40.jpg)
Quantizer (Quantization)
In general, if we have a (B+1)-bit binary two’s complement fraction of the form:
then its value is
The step size of the quantizer iswhere Xm is the full scale level of the A/D converter.The numerical relationship beween the code words and the quantizer samples is
Baaaa ...210◊
BBaaaa −−− ++++− 2...222 2
21
10
0
Bm
Bm XX 2/2/2 1 ==Δ +
][ˆ][ˆ nxXnx Bm=
![Page 41: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/41.jpg)
Example of quantization
![Page 42: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/42.jpg)
Analysis of quantization errors
Quantization errorIn general, for a (B+1)-bit quantizer with step size Δ, the quantization error satisfies that
when
If x[n] is outside this range, then the quantization error is larger in magnitude than Δ/2, and such samples are saidedto be clipped.
][][ˆ][ nxnxne −=
2/][2/ Δ≤<Δ− ne
)2/(][)2/( Δ−≤<Δ−− mm XnxX
![Page 43: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/43.jpg)
Analysis of quantization errors
Analyzing the quantization by introducing an error source and linearizing the system:
The model is equivalent to quantizer if we know e[n].
![Page 44: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/44.jpg)
Assumptions about e[n]
e[n] is a sample sequence of a stationary random process.e[n] is uncorrelated with the sequence x[n].The random variables of the error process e[n] are uncorrelated; i.e., the error is a white-noise process.The probability distribution of the error process is uniform over the range of quantization error (i.e., without being clipped).
The assumptions would not be justified. However, when the signal is a complicated signal (such as speech or music), the assumptions are more realistic.
Experiments have shown that, as the signal becomes more complicated, the measured correlation between the signal and thequantization error decreases, and the error also becomes uncorrelated.
![Page 45: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/45.jpg)
Example of quantization error
original signal
3-bit quantization result
3-bit quantization error
![Page 46: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/46.jpg)
Example of quantization error
8-bit quantization error
In a heuristic sense, the assumptions of the statistical model appear to be valid if the signal is sufficiently complex and the quantization steps are sufficiently small, so that the amplitude of the signal is likely to traverse many quantization steps from sample to sample.
![Page 47: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/47.jpg)
Quantization error analysis
2/][2/ Δ≤<Δ− nee[n] is a white noise sequence. The probability density function of e[n] is
![Page 48: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/48.jpg)
Quantization error analysis
The mean value of e[n] is zero, and its variance is
Since
For a (B+1)-bit quantizer with full-scale value Xm, the noise variance, or power, is
121 22/
2/
22 Δ=
Δ= ∫
Δ
Δ−
deeeσ
BmX
2=Δ
122 22
2 mB
eX−
=σ
![Page 49: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/49.jpg)
Quantization error analysis
A common measure of the amount of degradation of a signal by additive noise is the signal-to-noise ratio (SNR), defined as the ratio of signal variance (power) to noise variance. Expressed in decibels (dB), the SNR of a (B+1)-bit quantizer is
Hence, the SNR increases approximately 6dB for each bit added to the world length of the quantized samples.
⎟⎟⎠
⎞⎜⎜⎝
⎛−+=
⎟⎟⎠
⎞⎜⎜⎝
⎛ ⋅=⎟⎟
⎠
⎞⎜⎜⎝
⎛=
x
m
m
xB
e
x
XB
XSNR
σ
σσσ
10
2
22
102
2
10
log208.1002.6
212log10log10
![Page 50: Discrete-time Processing of Continuous-time Signals](https://reader030.fdocuments.net/reader030/viewer/2022020703/61fb350e2e268c58cd5b6dd8/html5/thumbnails/50.jpg)
Quantization error analysis
The equation can be further simplified for analysis. For example, if the signal amplitude has a Gaussian distribution, only 0.064 percent of the samples would have an amplitude greater than 4σx.Thus to avoid clipping the peaks of the signal (as is assumed inour statistical model), we might set the gain of filters and amplifiers preceding the A/D converter so that σx = Xm/4. Using this value of σx givesFor example, obtaining a SNR about 90-96 dB in high-quality music recording and playback requires 16-bit quantization.
But it should be remembered that such performance is obtained only if the input signal is carefully matched to the full-scale of the A/D converter.
dBBSNR 25.16 −≈