Adaptive decision feedback equalization with MLSE based on predicted signals

The authors introduce an adaptive channel equalization scheme which combines a maximum likelihood sequence estimator (MLSE) and a fractionally spaced decision-feedback equalizer (DFE). Besides compensating the channel distortion, the embedded DFE provides the MLSE detector with predicted values of the received signal so that the output detected symbols do not suffer any delay except the processing time. Both the bit error rate and the block error probability of the proposed MLSE-DFE scheme are significantly improved in comparison with the conventional DFE. The tradeoff between performance and complexity is controlled by a parameter m, representing the MLSE detection depth. To simplify the realization of the MLSE-DFE, a modified MLSE algorithm is also investigated. Analytical and simulation results are presented for uncoded quadrature phase shift keying (QPSK) signals over channels with severe distortions