| Knowledge of the channel is valuable for equalizer design. To estimate the channel, a training sequence, known to the transmitter and the receiver, is normally transmitted. However, transmission of a training sequence decreases the system throughput. Blind channel estimation uses only the statistics of the transmitted signal. Thus, it requires no training sequence, increasing the throughput.; Most communication systems employ some form of error-control code (ECC); however, most blind channel estimators ignore the code, and may fail at low SNR. Recently, blind estimators have been proposed that exploit the ECC and work well at low SNR. These algorithms are inspired by turbo equalizers and the expectation-maximization (EM) channel estimator.; In this research, we develop a low-complexity ECC-aware blind channel estimator. We first propose the extended-window (EW) algorithm, a channel estimator that is less complex than the EM estimator, and has better convergence properties. Furthermore, the EM algorithm uses the computationally complex forward-backward recursion (BCJR algorithm) for symbol estimation. With the EW estimator, any soft-output equalizer may be used, allowing for further complexity reduction.; We then propose the soft-feedback equalizer (SFE), a low-complexity soft-output equalizer that can use a priori information on the transmitted symbols, and is thus suitable for turbo equalization. Its coefficients are chosen to minimize the mean-squared error between the equalizer output and the transmitted symbols, and depend on the “quality” of the a priori information and the equalizer output. Simulation results show that the SFE may perform within 1 dB of a system using a BCJR equalizer, and outperforms other schemes of comparable complexity.; Finally, we show how the SFE and the EW algorithms may be combined to form the turbo estimator (TE), a linear-complexity ECC-aware blind channel estimator. We show that the TE performs close to systems with channel knowledge at low SNR, where ECC-ignorant channel estimators fail. |
