Keishi Murakami

A list-output Viterbi equalizer with two kinds of metric criteria

This paper proposes a list-output Viterbi equalizer (LVE) for wide-spread time-dispersive channels, which uses two kinds of metric criteria, in order to achieve a good trade-off between complexity and bit error rate (BER) performance. For the Viterbi equalization employing a state-reduction algorithm, the modified metric proposed by Ungerboeck (1974) is not always equivalent to the squared Euclidean distance. This paper proposes the following two LVEs: (1) to combine two kinds of metric criteria; (2) to select the metric criterion according to the channel impulse response. Finally, computer simulation shows that the proposed scheme improves the BER performance on frequency selective fading channels, even if the proposed scheme has smaller complexity than the conventional one.

A receiver with iterative soft decision and forward error correction

As a receiving method suitable for channels with intersymbol interference such as frequency selective fading, a method is proposed in which the soft decision using the hard decision sequence generated from the data after forward error correction (FEC) is alternated with FEC. In the conventional receiver with a serial configuration of the equalizer, deinverter, and decoder, the performance of the equalizer significantly affects the overall receiver characteristics. Hence, the performance of the equalizer is emphasized. On the other hand, in the proposed method the error generated in the initial equalization can be corrected by repetitive soft decision and FEC. Therefore, the method does not strongly depend on the performance of the equalizer used in the initial equalization. Hence, even if an equalizer requiring a smaller amount of process with a somewhat degraded performance is used in the initial equalization in place of a high-performance equalizer, the amount of process for the entire receiver can be decreased without degrading the characteristics of the proposed method very much. In this paper, the characteristics of the proposed method are evaluated by computer simulation. The following are the confirmed conclusions. (1) The proposed method using the maximum-likelihood sequence estimation (MLSE) as the equalization has superior BER characteristics over the conventional method using a maximum a posteriori (MAP) equalizer and MLSE. (2) In a transmission channel with a large delay spread, the degradation of the BER characteristics of the proposed method using the delayed decision-feedback sequence estimation (DDFSE) in place of MLSE is very slight while the amount of processing can be reduced substantially from that in the conventional method using MLSE.

An adaptive soft-output Viterbi equalizer for fast time-varying frequency-selective fading

In this paper, an adaptive soft-output Viterbi equalizer (SOVE) is proposed for fast time-varying frequency-selective fading. The demodulator can make a symbol-by-symbol or bit-by-bit soft-decision based on a posteriori probabilities. The proposed adaptive SOVE has the following advantages: (1) it has a superior tracking ability for fast time-varying frequency-selective fading, (2) it works in blind if the coding rules are suitably designed, and (3) a good BER is obtained at the output of the decoder by introducing interleaves. In this paper, these advantages are confirmed by computer simulations. In particular, in the presence of frequency-selective fading, we can show that the proposed method is superior to adaptive maximum-likelihood sequence estimation which carries equalizing and decoding simultaneously.