Soft decision decoding of block codes and concatenated block-convolutional codes using the stack algorithm

An approach to soft decision-decoding of block codes and concatenated block-convolutional codes is presented. These codes are shown to have a convenient tree structure that allows depth-first and metric-first sequential decoding techniques to be used to decode them. The unique features of the metric used for decoding are discussed, and a complete decoder of acceptable complexity is presented. For the additive white Gaussian noise (AWGN) channel, gains in excess of 1 dB at reasonable bit error rates (BER) with respect to conventional hard decision algebraic decoding are demonstrated for the (23,12) Golay code and for two Reed-Solomon codes.<
>