Blind SIMO channel estimation via a generalized MA estimation by spectral error minimization

This paper presents a generalized MA parameter estimation technique suitable for blind single-input multiple- output (SIMO) channel estimation. The method works by minimizing a cost function based on the spectral error obtained through the observed correlation sequences. The minimization is converted to a selection of a set of angles that is subsequently determined using a rapidly converging gradient based technique. The use of angles eases the optimization procedure due to the limited search space of plusmnpi. The estimates can be recursively updated, which is an attractive property for time varying SIMO channel identification. It is noted that the algorithm implementation has a very low complexity in matrix operations and mainly consists of angle manipulations which are of advantage in hardware realizations.