Exploiting input cyclostationarity for blind channel identification in OFDM systems

Transmitter-induced cyclostationarity has been explored previously as an alternative to fractional sampling and antenna array methods for blind identification of FIR communication channels. An interesting application of these ideas is in OFDM systems, which induce cyclostationarity due to the cyclic prefix. We develop a novel subspace approach for blind channel identification using cyclic correlations at the OFDM receiver. Even channels with equispaced unit circle zeros are identifiable in the presence of any nonzero length cyclic prefix with adequate block length. Simulations of the proposed channel estimator along with its performance in OFDM systems combined with impulse response shortening and Reed-Solomon coding are presented.