Near-Optimal Channel Estimation for OFDM in Fast-Fading Channels

In this paper, a low-complexity iterative receiver combining joint iterative channel estimation (CE) with symbol detection is proposed for coded orthogonal frequency-division multiplexing (OFDM) systems in fast-fading channels with Doppler frequencies up to 15% of the OFDM subcarrier spacing. The receiver exchanges information between the channel estimator and detector in an iterative fashion to obtain accurate estimates of the channel state information (CSI). The channel is modeled as a weighted sum of fixed basis expansion model (BEM) functions. The BEM coefficients are characterized as a multivariate autoregressive (AR) processes and estimated with a Kalman filter. The initial channel estimate is performed from sparse pilot signals. Data are detected and decoded, and the channel is estimated again based on the estimated transmitted data. The cycle of detection, decoding, and CE is repeated until convergence. It is shown that with a pilot-to-data ratio of 7/144 and with pilots consuming only 1/145 of transmission power, bit error rate (BER) performance within 0.1 dB of that achieved with perfect CSI is obtained for BER ≤ 10^-5.