基于深度学习的OFDM信道估计

针对多径正交频分复用(orthogonal frequency division multiplexing， OFDM)信道环境下信道频域选择性衰落导致下行链路信道估计性能受限的问题，提出一种基于深度学习的信道估计(deep learning-based channel estimation， DL-CE)方法。采用自回归过程对信道建模，利用深度学习设计信道估计网络追踪信道响应及其频域相关系数。通过迭代训练，基于深度学习的信道估计网络能够学习到自回归系数的最优估计，同时利用先验信道信息估计信道频域响应和频域相关系数。与传统方法相比，所提信道估计方法性能提升明显。

Channel estimation based on deep learning for OFDM systems

Aiming at the problem that downlink channel estimation performance is limited due to channel frequency domain selective fading in multipath Orthogonal Frequency Division Multiplexing (OFDM) channel environment, a deep learning-based channel estimation (DL-CE) method is proposed. The channel can be modeled by autoregressive process, and channel response and frequency domain correlation coefficients can be estimated by using the channel estimation network. Through iterative training, the deep learning based channel estimation network can learn the optimal estimation of autoregressive coefficients, and use the prior channel information to estimate the channel frequency domain response and the frequency domain correlation coefficients. Compared with the traditional methods, the performance of the proposed channel estimation method is significantly improved.