Cryptography Reference
In-Depth Information
Figure 11.16 - Performance of the MMSE turbo equalizer for BPSK transmission on
the Proakis C channel, with a 4-state rate
R
non-recursive non-systematic
convolutional code and a 16384 bit pseudo-random interleaver.
=1
/
2
Convergence of the iterative process occurs here at threshold signal to noise
ratio of 4 dB, and the turbo equalizer suppresses all the ISI beyond a signal to
noise ratio of 6 dB (after 10 iterations). Compared to the results obtained with
the MAP turbo equalizer (Figure 11.13), we can therefore make the following
remarks:
1. The convergence occurs later with MMSE turbo equalization (of the order
of 1 dB here, compared to MAP turbo equalization).
2. The MMSE turbo equalizer requires more iterations than the MAP turbo
equalizer to reach comparable error rates.
However, the MMSE turbo equalizer here shows its capacity to suppress all the
ISI when the signal to noise ratio is high enough, even on a channel that is known
to be dicult to equalize. It is therefore a serious alternative solution to the
MAP turbo equalizer when the latter cannot be used for reasons of complexity.
Second, the hypothesis of perfect knowledge of the channel parameters has
been removed and the turbo equalizer is simulated in the adaptive form, keep-
ing the same transmission parameters. The communication begins with the
transmission of an initial training sequence of 16384 symbols assumed to be per-
fectly known by the receiver. Then, frames composed of 1000 training symbols
followed by 16384 information symbols are periodically sent into the channel.
