A cost-effective maximum likelihood receiver for multicarrier systems

Equalization structures for maximum likelihood (ML) reception of data transmitted over intersymbol interference channels are studied. The equalizer that is best for the ML receiver is derived from a general theory of decision-aided equalization. The resulting optimum equalizers are linear and do not use previous decisions. If the equalizer complexity is permitted to be infinite, then a general optimum class of structures is derived that includes the decision feedback equalizer and the lesser-known autoregressive moving average filters. When a complexity constraint is also imposed on the equalizer, one of the structures in this class will be best for a given ML receiver. The best structure is found by a simple search procedure, which is given. The results indicate that near-optimum performance can be achieved by using this approach at a great computational reduction