Digital Signal Processing Reference
In-Depth Information
3.2
ISSR decoding for wideband QPSK
In the previously described signal reconstruction strategy, an exhaustive search
over all possible sequences was performed in order to find the symbol combi-
nation with the best resemblance to the spectral footprint of the received data
stream. Unfortunately, this approach turns out to be anything but efficient in
terms of processing power. The interference suppression and signal reconstruc-
tion (issr) decoder technique which will be introduced below does not rely
on standard equation-solving numeric algorithms, but is rather based on the
working principles of a nonlinear feedback system. Figure 3.3 shows the ba-
sic block diagram of the feedback system on which the issr decoder is based.
The forward path of the feedback loop embeds an
G/s
integration stage. The
closed-loop transfer characteristic of this system is given by a unity gain low-
pass filter of which the cut-off frequency is controlled by gain
G
. The input
of this low-pass filter is not a single scalar value, but a static vector formed by
the
n
consecutive qpsk samples from the received signal. Actually, the system
should be regarded as a parallel system of independent low-pass filters, each of
which is processing one sample from the static input vector. When the loop has
achieved convergence ([
]
0), exactly the same signal vector as was applied
to the input will be available at the output of the system. At this moment, it
should be clear that the potential benefits of this setup are somewhat limited.
=
static input vector
magnitude response H(s) [dB]
ε
+
G/s
in
1
out
1
-
in
n
out
n
0
H(s) = (1+s/G)
-1
G
frequency (s)
Figure 3.3.
The issr decoder implementation is based on a low-pass filter. The
feedback loop in the system tries to make the output vector equal to the
input signal. The speed at which this happens depends on gain factor
G
.
