Channel Estimation Algorithm Based on Deep Learning for OFDM System

In the process of channel estimation in orthogonal frequency division multiplexing (OFDM) systems, reliable channel estimation is the key task to achieve high data rate. The traditional channel interpolation algorithm is based on the assumption that the estimated values near the pilot are correlated. When the channel characteristics are discontinuous due to the time-varying and frequency-variable characteristics of the wireless channel, the estimated results will not be ideal. To solve this problem, this paper introduces a super-resolution reconstruction model, ESRGAN, to replace the interpolation processing in channel estimation. First, the estimated value is obtained by the least squares method (LS) at the pilot, and then the estimated value is compared to the pixel points in the low-resolution image. First, the channel characteristics are extracted through the convolution network, and then the mapping relationship is learned through multiple residual networks, and then amplified by the upper sampling layer. Finally, the estimation effect is continuously judged and improved through the discrimination network. The simulation results show that the channel estimation based on ESRGAN is better than the traditional LS algorithm and the same type of SRCNN model. And even if the model has fewer pilots, it also achieves excellent MSE performance, which shows great application potential in spectrum saving.