Linear cyclic correlation approaches for blind identification of FIR channels

Blind fractionally spaced equalizers rely on output only samples taken at a rate higher than the symbol rate in order to estimate the channel taps. Simple methods for mixed phase FIR channel identification are developed by solving a linear system of equations formed by sample cyclic correlations of the received oversampled data sequence. In addition to computational simplicity over existing multivariate alternatives, the resulting linear algorithm is easily modified to yield asymptotically minimum variance weighted least squares channel estimators. A “deterministic” approach is also developed which requires only persistently exciting inputs and leads to closed form data based solutions. Comparisons and tools for performance analysis are also described