Optimal pilot placement for channel tracking in OFDM

The problem of optimally placing pilot symbols for tracking frequency selective fading channels in an OFDM system is considered. The time-varying channel is approximated as a Gauss-Markov process. The Kalman filter, as the optimal tracking algorithm, is used for updating channel states in frequency. For a fixed percentage of pilot symbols, assuming pilot clusters are periodically sent through (L+1) OFDM subchannels for training, we optimize the cluster size by minimizing the average steady-state mean square error (MSE) of the channel estimator. It is shown that single pilot periodic placement achieves the minimum MSE.