Scribd
Upload
32 views3 pages

CT 214 Analog and Digital Communications Quantization: Assumption 1

1) Quantization error (qE) is reduced by choosing quantization intervals (Δk) to be smaller where the probability density function (f(x)) of the signal amplitude is higher, and larger where f(x) is lower. 2) If the signal amplitudes are uniformly distributed, the quantization intervals should be equal in length and qE is given by equation 2. 3) If the amplitudes are not uniform, qE can be minimized by choosing the quantization levels (yk) and intervals (Δk) such that the relation in equation 3 is satisfied, where f(yk) is the probability density at the quantization level and C/N is a constant.

Uploaded by

diptanshujain
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
32 views3 pages

CT 214 Analog and Digital Communications Quantization: Assumption 1

1) Quantization error (qE) is reduced by choosing quantization intervals (Δk) to be smaller where the probability density function (f(x)) of the signal amplitude is higher, and larger where f(x) is lower. 2) If the signal amplitudes are uniformly distributed, the quantization intervals should be equal in length and qE is given by equation 2. 3) If the amplitudes are not uniform, qE can be minimized by choosing the quantization levels (yk) and intervals (Δk) such that the relation in equation 3 is satisfied, where f(yk) is the probability density at the quantization level and C/N is a constant.

Uploaded by

diptanshujain
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

CT 214 Analog and Digital Communications

Quantization

• Assumption 1 : Assume that the signal s(t) takes values in the interval ±V volts.
– Recall- The quantization error
N
X ∆2 k
qE = pk , (1)
12
k=1

where

N = Number of quantizing intervals


∆k = length of k th interval
pk = probability of signal value being in the k th interval

1. Let all ∆0k s be the same -uniform then,


N
∆2k X ∆2
qE = pk = k , and
12 12
k=1
2V
∆ =
N
1 4V2 1 V2
qE = = (2)
12 N 2 3 N2
• We take note of two factors:
1. If signal amplitudes are uniformly distributed in (−V, +V ), then there is no reason for the
∆0k s to be unequal. In this case ∆ = 2V
N , and result (2) will hold.
PN ∆2
2. If signal amplitudes are not uniformly distributed, then eq(1), qE = k=1 12k pk , suggests
that it might be possible to reduce qE by choosing ∆k smaller where pk is higher and
vice-versa. Panter and Dite showed one way of doing this as follows.
• Assumption 2: Let the signal take values in (x, x + dx) with probability f (x)dx , where
f (x) is the probability density function for signal amplitude.

• Property 1:
ZV
f (x) dx = 1, f (x) ≥ 0 for ∀x
−V

1
Figure 1: The figure shows probability density function f (x) of a typical signal.

• Assumption 3: Quantizing intervals are small enough to approximate f (x) in a given interval
by the value of f (x) at the center of the interval (see Fig. 1 ) and the quantizing level yk is the
midpoint of the interval. These can be considered as reasonable approximations.
– Then eqn (1) can be written as
X ∆2 X ∆2
k k
qE = pk = f (yk )∆k
12 12
k k
X ∆3k
= f (yk )
12
k
P
– We also require k ∆k = 2V .
– substituting
1
f 3 (yk )∆k = µk ,
Then
N
X
qE = µ3k − To minimize qE
k=1

– We determine the constraint as follows :


ZV
1 1
f (yk )∆k ∼
X X
µk = 3 = f 3 (x) dx
k k −V

– Let their integral = C , a constant.


– So, the problem can be stated as :
N
X
Minimize eq = µ3k given that
k=1
PN
– µk = C. As there are no weightages to µ0k s in qE .
k=1
PN
– and the constraint is k=1 µk = C constant, all µ0k s must be equal. Hence
N
X
µk = N µ = C and µ = C/N
k=1

2
and qE is minimum, when {yk } and {∆k } are chosen such that
1
f 3 (yk )∆k = C/N (3)

– Also, {yk } are midpoints of {∆k }.

You might also like