An efficient new technique for accurate bit error probability estimation of ZJ decoders

Sequential decoding is a very powerful decoding procedure for convolutional codes which is characterized as a sequential search for the correct transmitted path through a large decision tree with random node metrics. The authors present a new efficient technique for accurately estimating the bit error rates (BERs) of stack or Zigangirov-Jelinek (ZJ) sequential decoders for specific time-invariant convolutional codes. In contrast to a brute-force simulation approach, this method achieves computational efficiency by using a mean-translation importance sampling biasing method along with a set of easily checked rules which immediately eliminate the need of running the computationally intensive ZJ algorithm in a large number of simulation trials