A structured channel estimator for maximum likelihood sequence detection in multipath fading channels

This paper describes a channel estimator using known prior information about the transmit and receive filters, it is shown that the composite channel lies in a certain subspace obtained from the impulse responses of these filters. A structured linear channel model is then developed that is linearly parameterized by an unknown vector. To illustrate the potential usefulness of such an approach, the estimated structured channel is used in a multisensor and oversampled maximum likelihood sequence estimation (MLSE) receiver. We also present expressions on the pairwise error probability for the MLSE receiver based on the structured channel model. Using these results, we investigate the phenomenon of error flooring.