Optimal blind equalization and symbol error analysis of fractionally-sampled channels

Blind fractionally-spaced equalizers employ the second-order cyclostationary statistics of the channel output in order to derive both zero-forcing and MMSE solutions. In this work, optimal minimum variance estimates for the zero-forcing and MMSE equalizers are developed. Performance analysis is carried out by deriving symbol error probabilities and closed form expressions for the variance of the equalizer tap estimates. In addition, simulations are presented in order to demonstrate the performance of these equalizers and to corroborate the theoretical symbol error expressions.