Burst mode non-causal decision-feedback equalisation and blind MLSE

We propose a decision-aided equalizer that uses past symbol decisions, like classical decision-feedback equalizers, but also future decisions, hence the name non-causal decision-feedback equalizer (NCDFE). Its derivation is done in a burst processing and multichannel context, using maximum likelihood sequence estimation (MLSE). The forward filter of the NCDFE is proportional to the channel matched filter and the feedback filter is the cascade of the channel and the forward filter (without the central coefficient). This NCDFE corresponds to the unbiased MMSE solution. Blind MLSE is usually solved using alternating minimizations. We propose to replace the repeated use of the Viterbi algorithm with the much less complex NCDFE. Finally, some simulation results are presented