An adaptive threshold strategy for soft decision Viterbi decoding

Soft decision decoding of convolutional codes with analog inputs is known to have an improvement of about 2 dB over hard decision decoding. Compared to the analog soft decision decoders, decoders that use uniformly quantized receiver outputs exhibit performance degradation, but they still outperform the hard decision decoders. In this paper, to close the performance gap between analog and equispaced quantized inputs soft decision decoders, two ideas are examined: (i) nonuniform spacing in quantizer thresholds; and (ii) adaptability in threshold levels according to signal-to-noise ratio (SNR) conditions in the channel. An adaptive soft-decision quantizer with quantization thresholds controlled by the automatic gain control unit is investigated and demonstrated to improve the performance of a soft-decision Viterbi decoder in communication systems subject to the additive white Gaussian noise (AWGN) and Rayleigh fading.