高移动性Jakes信道的学习与估计

在高移动场景下, 信道具有快速时变性和非平稳特性, 对信道的准确估计提出了新的挑战。针对高移动性Jakes信道, 提出一种基于图像重建和恢复原理的信道学习估计网络。首先，根据Jakes信道矩阵中局部相关特性, 构建快速超分辨卷积神经网络提取信道特征, 并对信道插值完成信道图像建模。然后, 利用去噪神经网络降低信道噪音的影响, 进一步提高估计精度。最后，通过时域和频域的仿真测试, 所提方案估计性能优于传统算法。在与现有基于深度学习最新方法比较中, 所提方案也有性能优势, 并且收敛速度更快。

Learning and estimation of high mobility Jakes channel

In high mobility scenarios,the channel has the characteristics of fast time-varying and non-stationary,which poses a new challenge to the accurate channel estimation.For high mobility Jakes channel,a channel learning and estimation network based on image reconstruction and recovery principle is proposed.Firstly,according to the local correlation characteristics of Jakes channel matrix,a fast super-resolution convolution neural network is constructed to extract channel features,and channel image modeling is completed by channel interpolation.Then,the denoising neural network is used to reduce the influence of channel noise and further improve the estimation accuracy.Finally,the simulation results in time domain and frequency domain show that the proposed scheme performs better than the traditional algorithm.Compared with the latest methods based on deep learning,the proposed scheme also has performance advantages and faster convergence speed.