0.64 0.66 0.68 0.7 0.72 0.74
Classical LFSR approach
0.64 0.66 0.68 0.7 0.72 0.74 0.76
Multiple LFSR approach
0.66 0.68 0.7 0.72 0.74
Leap Ahead LFSR approach
Figure 2.6: Histogram of three approaches of LFSR–based SNG with the
expected probability 0.7
Under such configuration, it can be shown in
Fig.2.6 that from left–to–right
there are no considerable differences in the distributions of three
approaches around central value 0.7. The same conclusion when we
consider the Empirical CDF function corresponding to each
approach as shown in Fig.2.7.
Moreover, Table 2.1 also shows a
similar quality for all three method when the error is evaluated.
Nevertheless, due to the results in this table, with a strict
requirement in generating SN, Leap–Ahead architecture seems to
be the best one playing as SNG based on LFSR.
In addition, it is theoretical to see that the classical LFSR
method is not amenable for generating SN in parallel. Hence, we can
conclude that with the nearly same quality of SN generation,
multiple LFSRs and Leap–Ahead LFSR approaches are the
good choices. However, using multiple LFSRs deals with the complexity in
harware design, while Leap–Ahead architecture has only one LFSR.
Consequently,
Leap–Ahead approach consumes less harware material than
Multiple LFSRs one. Furthermore, Multi- ple LFSRs design
has 16×32 cells need to be initialised at the beginning, while
Leap–Ahead architecture
has only 32 cells, i.e, there are more slices for Multiple LFSRs
method in innitial phase when implement
on FPGA platform. In conclusion, Leap–Ahead architecture will be
a better selection than multiple LFSRs one not only in quality
aspect, but also in low–cost evaluation.