A frequently used pictorial representation of the convolutional encoding
and decoding process is the trellis diagram. At each stage in
the trellis, a number of valid symbols or bit patterns exist, and over time, the decoder
builds up a picture of the path exhibited by the received data through the trellis. With soft decision decoding, each element in the path can be weighted with a confidence factor using the 'side' information. The job of the decoder is, in effect, to check all possible paths through the trellis (not just the most apparent one) and add up all the weights over N sections, to see which path in fact gives the highest confidence rating. This path is then chosen as the correct set of received symbols and is decoded to give the output bit stream. |
![]() |
The Viterbi algorithm is very effective at performing this path search process and in recent years has been effectively fabricated on silicon. More information on trellis coding and the Viterbi algorithm can be found in Ziemer and Peterson (1992). |