A Low Complexity Learning-Based Channel Estimation for OFDM Systems With Online Training

In this paper, we devise a highly efficient machine learning-based channel estimation for orthogonal frequency division multiplexing (OFDM) systems, in which the training of estimator is performed online. A simple learning module employed proposed estimator. The process thus much faster and required data reduced significantly. Besides, construction approach utilizing least square (LS) results so that can be collected during transmission. feasibility novel verified by theoretical analysis simulations. Based on approach, two alternative generation schemes are proposed. One scheme transmits additional block pilot symbols to create data, while other adopts decision-directed method does not require extra overhead. Simulation show robustness method. Furthermore, shows better adaptation practical imperfections compared with conventional minimum mean-square error (MMSE) estimation. It outperforms existing techniques under varying conditions.