ML blind channel estimation in OFDM using cyclostationarity and spectral factorization

Channel estimation is vital in OFDM systems for efficient data recovery. In this paper, we propose a blind algorithm for channel estimation that is based on the assumption that the transmitted data in an OFDM system is Gaussian (by central limit arguments). The channel estimate can then be obtained by maximizing the output likelihood function. Unfortunately, the likelihood function turns out to be multi-modal and thus finding the global maxima is challenging. We rely on spectral factorization and the cyclostationarity of the output to obtain the correct channel zeros. The Genetic algorithm is then used to fine tune the obtained solution.