Adaptive Viterbi decoding of turbo codes with short frames

Turbo codes can significantly improve the performance of a communications system when long frames are used. However, their performance degrades when the frame length decreases. Furthermore, decoding turbo codes with the MAP algorithm or the SOVA is quite complex. The adaptive Viterbi algorithm (AVA) has been proposed as an alternative to the Viterbi algorithm for convolutional codes. It requires a smaller average number of survivors while yielding a bit error probability very close to that of the Viterbi algorithm. In this paper, it will be shown that the soft-output adaptive Viterbi algorithm provides the same advantages with turbo codes. Hence, codes with a longer memory can be used to improve the error performance of short frames and lower the error floor.