Data-Aided Kalman Tracking for Channel Estimation in Doppler-Affected OFDM Systems

This paper considers a pilot-aided/data-aided Kalman channel estimator for OFDM in fast time-varying (TV) channels. Capitalizing on a basis expansion model (BEM) and on a frequency-domain estimation philosophy, the OFDM system is designed to periodically switch between a pilot-aided and a data-aided mode in order to reduce the rate penalty introduced by the training pilots. The suggested philosophy is effective in tracking the channel changes in the data-aided mode with a negligible channel mean squared error (MSE) penalty, if the Kalman filter prediction capability is coupled with an iterative data-aided estimation that is equipped by an opportune selection of the detected data. Appropriate data selection metrics for LMMSE and DFE equalizers are also provided, and the impact of the proposed channel estimation on the ultimate BER performance of each equalizer is investigated by simulations