一种新型的基于深度学习的时变信道预测方法

A novel deep learning based time-varying channel prediction method

针对高速移动正交频分复用系统,提出了一种新型的基于深度学习的时变信道预测方法。为了避免网络参数随机初始化造成的影响,本文方法首先基于数据与导频信息获取较理想的信道估计,利用其对BP神经网络进行预训练处理,以获取理想的网络初始参数;然后,基于预训练获取网络初始值,利用基于导频获取的信道估计对BP神经网络进行再次训练,以获取最终的信道预测网络模型;最后,本文方法基于该预测网络模型通过线上预测实现了时变信道的单时刻与多时刻预测。仿真结果表明,本文方法可以显著地提高时变信道预测精度,且具有较低的计算复杂度。

For high-speed mobile orthogonal frequency division multiplexing(OFDM)systems,a novel time-varying channel prediction method based on deep learning was proposed.To avoid the influence caused by random initialization of network parameters,the proposed method firstly obtains an ideal channel estimation based on data and pilot,and then pre-trains the back propagation(BP)neural network based on the channel estimation to obtain the ideal network initial parameters.Then,based on the initial network value obtained by pre-training,the proposed method uses the channel estimation based on pilot to train the BP neural network again.Finally,the proposed method realizes the single-time-and multi-time prediction of time-varying channels through on-line prediction.Simulation results show that the proposed method can significantly improve the prediction accuracy of time-varying channels and has a low computational complexity.