Blind equalization in the presence of jammers and unknown noise: Solutions based on second-order cyclostationary statistics

This correspondence addresses the blind identification of a linear time-invariant channel using some second-order cyclostationnary statistics. In contrast to other contributions, the case where the second-order statistics of the noise and of the jammers are totally unknown is considered, It is shown that the channel can be identified consistently by adapting the so-called subspace method of Moulines et at This adaptation is valid for fractionally spaced systems and, more interestingly, for the general systems exhibiting transmitter induced cyclostationnarity introduced by Tsatsanis and Giannakis. The new subspace method is based in both cases on a common tool, i.e., a general spectral factorization algorithm, The identifiability conditions are specified and some simulation examples are given.