Abstract—Symbol spaced blind channel estimation methods
are presented which can essentially use the results of any existing
blind equalization method to provide a blind channel estimate
of the channel. Blind equalizer’s task is reduced to only phase
equalization (or identification) as the channel autocorrelation is
used to obtain the amplitude response of the channel. Hence, when
coupled with simple algorithms such as the constant modulus
algorithm (CMA) these methods at baud rate processing provide
alternatives to blind channel estimation algorithms that use
explicit higher order statistics (HOS) or second-order statistics
(subspace) based fractionally-spaced/multichannel algorithms.
The proposed methods use finite impulse response (FIR) filter
linear receiver equalizer or matched filter receiver based infinite
impulse response+FIR linear cascade equalizer configurations to
obtain blind channel estimates. It is shown that the utilization of
channel autocorrelation information together with blind phase
identification of the CMA is very effective to obtain blind channel
estimation. The idea of combining estimated channel autocorrelation
with blind phase estimation can further be extended to
improve the HOS based blind channel estimators in a way that
the quality of estimates are improved.