Performance of sub-optimal normalisation schemes for a turbo decoder using the soft output Viterbi algorithm

It has been shown that the extrinsic information produced by the soft output Viterbi algorithm (SOVA) is too optimistic. To compensate for this, the extrinsic information must be normalised. Standard normalisation techniques involve calculating the mean and variance of all of the bits in the SOVA output frame. Since this adds complexity to every iteration when the SOVA is used in a turbo decoder, there is merit in investigating the performance of reduced complexity techniques. We present the performance of a parallel concatenated SOVA based turbo decoder using the suggested reduced complexity normalisation techniques. BER results are presented for the case of an AWGN channel.