Skip to Main Content
The iterative soft output Viterbi algorithm (SOVA) is a sub-optimum algorithm when it is used to decode turbo codes. By normalizing its extrinsic information we get a performance improvement compared to the standard SOVA. In particular, if the extrinsic information is increased in the last decoding iteration, an additional coding gain improvement is noticed. For example, this is 0.25 dB for a frame length of 1000 bits in the additive white Gaussian noise (AWGN) channel as well as in an uncorrelated Rician fading channel at bit error rate (BER) of 10-6. Also, this normalized SOVA is only about 0.25 dB worse than a turbo decoder using the log-MAP algorithm, both in the AWGN channel and in an uncorrelated Rayleigh fading channel at BER of around 10-6. Furthermore, with an 8-state component code, a frame length of 500 bits performs 0.125 dB better than a 16-state bidirectional (bi)-SOVA turbo decoder at BER of 10-5 in the AWGN channel.