Improved OFDM channel estimation based on transform domain de-noising

Conventional DFT-based channel estimations improve the estimators by suppressing time domain noise. However, the performance will not be very satisfactory for the presence of the residual noise. In tunnel environment, it will be worse due to abundant noise and multipath. In this paper, an improved channel estimation algorithm is proposed based on quadratic transform domain de-noising. In each filtering process, we set noise threshold which varies along with the signal-to-noise ratio dynamically. Simulation results show that the proposed estimator outperforms the conventional DFT-based channel estimator on both MSE and BER. Furthermore, the algorithm´s low complexity brings the advantage of easy implementation for the estimators.