| miracles 2007-03-30, 7:11 pm |
| The simulation setup is quite simple, however results are really
strange.
One thing needed to mention is that a perfect interleaver is used in
the following two situations. And matlab build-in function 'convenc'
and 'vitdec' are used for convolutional encoding and viterbi decoding.
First, a long packet (10^6 bits) is convolutionally encoded and
transmitted through the Rayleigh fading channel. BER is calculated
versus Eb/No.
Second, a short packet (100 bits) is convolutionally encoded and
transmitted through a Rayleigh fading channel. And 10000 such packets
are transmitted. And BER is calculated and averaged for those
packets.
However, results for the two situations are not the same. If BER is
plotted versus Eb/No, there is actually a small gap between two
curves. Since perfect interleaver is used in two cases, the result
should be the same, right?
Do you have any idea about the problem?
thanks.
|