Iterative MAP Multi-User OFDM over Rapidly-Varying Frequency-Selective Channels

In this paper, we introduce an Expectation Maximization (EM) based iterative multi-user OFDM detector, the detector works in tandem with an a posteriori probability (APP) channel decoder to form a Turbo processor, suitable for use in a rapidly changing multi- path fading environment. The system uses spatial diversity at the receiver, by employing a number of uncorrected (independent) antennas. The normalized Doppler frequencies are in the order of 0.02 to 0.04, i.e. fast changing channel conditions. The system uses a single pilot slot every 10 - 15 slots, for initial channel estimation (acquisition) per user, so only minimal overhead is required. We show that this configuration achieves very high gains first by virtue of the EM detector that simultaneously detects user data and tracks the channel changes, and second by employing the APP decoder, while at the same time it converges very closely to the ideal, perfect channel side information receiver.