An Improved VA and Its Application in Turbo Code Decoding

This paper proposes an improved Viterbi algorithm (VA) and its application in soft-output Viterbi algorithm (SOVA) for turbo code decoding. The modifications target the decoded state error. According to the simulation results, for Viterbi decoder, the decoding performance is equivalent to that of conventional algorithm at twice of decoding depth, when the decoding depth delta is approximate to 2-3 times of memory length m in encoder. Thus the size and consumption can be reduced by the modification. With the survivor memory unit (SMU), the size and power consumption can be reduced about 20%, and with the trace back unit (TBU), about 30%. For SOVA-based turbo decoder with 5 iterations, the performance of modification with delta = 12 is better than that of conventional decoder with delta = 20. For the size and power consumption of SMU and TBU, the reduction ration can be about 20%.