Cryptography Reference
In-Depth Information
1.00E+00
LDPC (5
0 it)
BP
Amin*
lambda=4
lambda=3
lambda=2
BPBased
1.00E01
1.00E02
1.00E03
1.00E04
1.00E05
1.00E06
1
1.5
2
2.5
3
3.5
4
4.5
5
Eb/No
Figure 9.17 - Comparison of performance between the
3
−
min
and
A−
min
∗
algorithms
in the case of decoding an irregular code
C
3
.
and we pass from a quantified scalar
a
q
to a non-quantified scalar
a
by the
relations:
a
q
=
trunc
a
2
nq−
1
δ
+0
.
5
,
(9.39)
δ
2
n
q
−
1
a
=
a
q
where
trunc
designates the truncature operation.
These two parameters can influence the decoding performance. Too low a dy-
namic allows an error floor to appear on the error rate curves. This floor appears
much earlier than that associated with the minimum distance of the code.
However it is important to note that increasing the dynamic without increas-
ing the number of quantization bits increases the value of the bit with the lowest
weight, and consequently decreases the precision of the computations done in
the PNP. Increasing the dynamic without increasing the number of bits degrades
the decoding performance in the convergence zone.
