【摘要】 为了提高海上正交频分复用(OFDM)信道估计的准确性，提出了一种基于时序特征融合的神经网络信道估计方法。该方法在卷积网络(CNN)的基础上融入了时序特征融合模块，以增强特征提取能力，提高了信道估计的准确性。时序特征融合模块通过双向长短期记忆网络(BiLSTM)提取高级时序特征，实现了基本特征与高级时序特征的融合，并使用海上实测数据集进行训练和验证。实验结果表明，所提方法的误码率比传统信道估计方法有3db左右的提升，比全连接网络的信道估计方法有2db左右的提升，在海上通信环境中表现出良好的信道估计性能。更多还原

【Abstract】 In order to improve the accuracy of OFDM channel estimation in maritime environments, a neural network-based channel estimation method is proposed, leveraging the fusion of temporal features. This approach introduces a temporal feature fusion module on the foundation of Convolutional Neural Networks(CNN) to enhance feature extraction capabilities, thereby improving the accuracy of channel estimation in maritime environments. The temporal feature fusion module utilizes a Bidirectional Long Short-Term Memory network(BiLSTM) module to extract advanced temporal features, achieving the fusion of fundamental features with high-level temporal characteristics. The model is trained and validated using actual maritime measurement data. The experimental results show that the bit error rate of the proposed method is about 3db higher than that of the traditional channel estimation method and about 2db higher than that of the channel estimation method of the fully connected network, which shows good performance in the Marine communication environment.更多还原