Digital Signal Processing Reference
In-Depth Information
Figure
20.7
Typical
difference
between
the
(a)
traditional
and
(b)
suggested
data
compression/decompression schemes
The advantage of the considered pseudo-randomized data compression
approach over the conventionally used data compression techniques is illustrated
in Figure 20.7(b). The functional block for data compression is not included in the
second scheme. It is not needed as data in the second case are actually presented
in a compressed form at the output of the digitizer. Nothing special has to be done
to compress the acquired data. Therefore it is evident that designs of the data ac-
quisition subsystems exploiting this approach to data compression could be much
simpler with reduced power consumption. The full computational burden in this
case is placed on the functional block performing data decompression. For that,
the waveform of the original signal usually has to be reconstructed. Once that
is accomplished, this waveform could be represented by its equidistant sample
values taken at a sufficiently high sampling rate to meet the requirement of the
sampling theorem.
The embedded systems if they are built according to the scheme given in
Figure 20.1 would support this data compression and reconstruction scheme.
When they are used for dealing with analog signals, the data representing them
are compressed at the stage of signal digitizing. In cases where these data are
immediately processed, the computer performs the processing as required. If
the given data have to be converted into a digital periodic signal sample value
sequence, the computer performs data decompression and the signal waveform
is reconstructed for that.
The computer of a particular embedded system apparently can reconstruct
signal waveforms either by processing the digital signal taken from the output
of the pseudo-randomized ADC included into the structure of this embedded
system or by processing some other external digital signal encoded in a similar
way. Figure 20.8 illustrates a few versions of the digital signals represented in
various forms. The basic digital signal taken off the digitizer output is a sequence
of the input signal sample values obtained at nonuniformly spaced time instants
