Improved bit error rate performance of convolutional codes with synchronization errors

Abstract

A maximum a posteriori decoder is given for convolutional codes when transmitted over a binary random error channel that exhibits insertion, deletion and substitution errors. It is shown that the error-correcting performance of these codes can be improved by more than two orders of magnitude when compared to similar systems found in the literature, for the same decoding complexity. This is achieved by the simple addition of a pseudo-random sequence at the output of the encoder. The complexity of the decoder is limited by reducing the size of its state space using probabilistic arguments, avoiding the use of experimental trial-and-error techniques used in similar decoders.