Digital Signal Processing Reference
In-Depth Information
probability by reconstructing the paths. This is precisely the principle of
operation of the so-called Viterbi decoder, named after its inventor. The
Viterbi decoder is virtually the counterpart of the convolutional decoder
and there is, therefore, no convolutional decoder. The Viterbi decoder is
also much more complex than the convolutional coder.
t0
t1
t2
t3
0
/
00
0
/
00
0/00
00
0/11
1
/
1
1
1
/
11
01
0/10
1/00
10
1/01
11
States
of
shift
register
time
Fig. 14.16.
Trellis diagram
After the convolutional coding, the data stream is now inflated by a fac-
tor of 2. For example, 10 Mbit/s have now become 20 Mbit/s but the two
output data streams together now carry 100% overhead, i.e. error protec-
tion. On the other hand, this correspondingly lowers the net data rate avail-
able. This overhead, and thus also the error protection, can be controlled in
the puncturing unit (Fig. 14.17.), e.g. the data rate can be lowered again by
selectively omitting bits. The omitting, i.e. the puncturing, is done in ac-
cordance with an arrangement called the puncturing pattern, which is
known to the transmitter and the receiver.
This makes it possible to vary the code rate between 1/2 and 7/8. 1/2
means no puncturing, i.e. maximum error protection, and 7/8 means mini-
mum error protection and a maximum net data rate follows correspond-
ingly. At the receiving end, punctured bits are filled up with 'Don't Care'
bits and are treated like errors in the Viterbi decoder and thus recon-
structed. Up to here the processing stages of DVB-S and DVB-T are 100%
alike. In the case of DVB-T, the two data streams are combined to form a
common data stream by alternately accessing the upper and lower punc-
tured data stream. In DVB-S, the upper data stream and the lower data
stream in each case run directly into the mapper where the two data
Search WWH ::
Custom Search