Improved decoding with the SOVA in a parallel concatenated (Turbo-code) scheme

We show that the soft output of the soft-output Viterbi-decoder (SOVA) suffers from two distortions. Firstly, for bad channels the reliability information of the decoder output is too optimistic. The output can be considered as being multiplied by a factor, that depends on the current bit-error-rate (BER). To become closer to the true log-likelihood ratio the output has to be normalized. Secondly, the soft-output of the SOVA-when used in Turbo decoding-is effected by a correlation between the so called extrinsic and intrinsic information. Since the extrinsic information is fed forward to be the a-priori information in the next decoding stage and is treated as being uncorrelated with the systematic-information, a correcting term has to be introduced in order to compensate this correlation. We show how these distortions can be corrected and we apply the correcting measures on the SOVA that is used as component decoder of a Turbo-code scheme. Finally, we provide simulation results that illustrate the gain due to normalizing the extrinsic information and exploiting the correlation in the soft output