Decision-directed tracking for multi-user broadband mobile OFDM

This paper presents two new multi-user techniques that aid in the decision-directed tracking of a rapidly time-varying OFDM channel. The first is based on applying gradient optimization techniques to a least squares multi-user frequency-domain channel estimation problem. The second technique is based on a method for modeling a time-varying frequency response as a linear combination of a small number of time-invariant “Doppler channels.” This technique provides a formula for updating these Doppler channels, which greatly improves the performance of decision-directed tracking. These techniques are tested for different decision-directed methods in a simulated 6.0 MHz OFDM system with a 330 Hz maximum Doppler frequency and an 8.0 μsec maximum delay spread