Semi-blind channel identification in OFDM

Semi-blind channel estimators promise to dramatically reduce the number of pilots needed for accurate channel estimation over pilot-only based techniques while providing much better performance than blind algorithms. This paper presents a novel semi-blind multi-user channel estimator for OFDM based on the maximum likelihood estimator. The estimator works by iterating between estimating the channels between multiple transmit antennas and each receive antenna and the respective symbols for each transmitter. Simulation results show that the semi-blind channel estimator is effective even when the frequency-domain pilot spacing is greater than the Nyquist spacing.