Digital Signal Processing Reference
In-Depth Information
difference between the current sample and the previous sample; with such an
approach, a smaller number of bits is required for encoding. The disadvantage of
using this type of differential quantization is that a memory is needed to maintain
knowledge of the previous sample. A slightly different approach involves com-
paring the current sample to its predicted value (determined from previous sam-
ples) rather than the previous sample, then encoding the difference. The encoding
here is called differential PCM (DPCM). The disadvantage of DPCM is that it
needs a feedback loop and associated memory.
One way to achieve DPCM is the delta modulation (DM) system shown in
Fig.
3.31
. This DM is a first-order analog feedback system that consists of a
quantizer clocked at the sampling rate f
s
and an integrator (I). The integrator does a
linear prediction x
ð
n
Þ
of the current sample x(n), based on the previous sample
x(n - 1) and the most recent error. The quantized output is a comparator that
gives either 1 (logic-1) if x
ð
n
Þ
x
ð
n
Þ
and -1 (logic-0) if x
ð
n
Þ
\x
ð
n
Þ:
Only a 1-bit
word is needed for encoding, since the output is either logic-0 or logic-1. The
integrator has a scaling factor D
;
which is referred to as the step-size.
The operation of the DM can be further explored with an example. Consider the
input analog signal shown as a dotted curve in Fig.
3.32
. tisx
ð
t
Þ¼
0
:
8 sin
½
2p
ð
0
:
1
Þ
t
þ
0
:
1
volt. The sampling frequency is f
s
= 10 Hz (i.e., T
s
= 0.1
Clock (
f
s
)
e
(
t
)
1
−1
∼
x
(
t
)
x
(
t
)
y
(
t
)
Δ
∫
LPF
−1
Quantizer
Integrator
^
x
(
t
)
[ DM Decoder ]
Δ
∫
y(t).dt
[ estimate of
x
(
t
) ]
[ DM Encoder ]
Fig. 3.31
Delta modulator and demodulator
T
s
1
kT
s
−1
k
0 1 2 3 4 5
30
Fig. 3.32
Delta modulation waveforms
Search WWH ::
Custom Search