SPC10-1: Fractionally-Spaced a-EM-MLSE Receivers Over Frequency-Selective Fading Channels

The EM (expectation-maximization) algorithm is used to solve the decision delay problem existed in maximum- likelihood sequence estimation (MLSE) receivers, while the a- EM-MLSE receivers provide an alternative controllable convex divergence scheme for accelerating the overall convergence speed at the expense of more complicated mathematical structures. Instead of the conventional symbol-spaced (SS) MLSE structure, a new design based upon T/2-spaced samples for a-EM MLSE receivers has been scrutinized. Simulation results reveal that the number of branch metric computations and the accuracy of received sampling data dominate the overall convergence speed. That is, fractionally-spaced (FS) adaptive a -EM MLSE receivers have not only demonstrated a better BER performance, but they also could further speed up the convergence speed for a ges - 0.5 under SNR < 20 dB. Furthermore, the FS adaptive a-EM-MLSE receiver gives the more accurate sampling data for compensating the penalty of its twice number of branch metric computations so that both FS and SS receivers would reach almost the same convergence speed under the circumstance of SNR > 20 dB and a ges - 0.5.