A Fractional Doppler Channel Estimation Method for OTFS System Based on DCNN
When OTFS modulation is used for actual data transmission,the Doppler resolution is usually low,leading to fractional Doppler frequency shifts and causing Doppler inter-symbol interference in fading channels,which reduces the accuracy of channel estimation.To address this issue,a deep learning-based fractional Doppler channel estimation method for OTFS systems is proposed.Firstly,the cross-correlation algorithm is used for the initial estimation of fractional Doppler channels in the method;and then,a deep convolutional neural network is built and trained to optimize the preliminary channel estimation results,thus effectively improving the accuracy of OTFS fractional Doppler channel estimation.Simulation experiments show that by combining the advantages of traditional algorithms and deep learning,the proposed method achieves an approximate 6 dB performance gain,effectively enhancing the accuracy of OTFS fractional Doppler channel estimation.Moreover,the method is capable of effectively addressing channel mismatch,with performance differences in various high-mobility scenarios being less than 30%,demonstrating a certain level of robustness.
NETL
NSTL
万方数据
