基于深度展开神经网络的融合感知通信场景信道估计

Channel Estimation for Integrating Sensing and Communications Based on Deep Unfolding Neural Network

为了提高毫米波大规模MIMO系统中信道估计的准确性,将雷达感知与通信系统融合,提出一种基于深度展开神经网络的雷达辅助通信信道估计方法。该方法首先利用雷达的回波信号获取目标的角度,并将角度信息传输给通信模块。在通信模块中,将信道估计问题转化为稀疏信号恢复问题。利用感知获得的角度信息作为先验信息,设计了一种新的软阈值收缩函数,并提出了一种先验信息辅助的深度展开信道估计神经网络。仿真结果表明,提出的信道估计方法相较于现有的压缩感知类方法和深度学习类方法具有更低的信道估计误差。

To enhance the accuracy of channel estimation in mmWave Massive MIMO systems,a radar-assisted communication channel estimation method based on deep unfolding neural networks is proposed,integrating radar sensing with the communication system.Initially,radar echo signals are employed to extract target angles,which are subsequently conveyed to the communication module.In the communication module,the channel estimation problem is transformed into a sparse signal recovery challenge.Capitalizing on the angle information obtained through sensing as prior knowledge,a novel softthresholding shrinkage function is devised.Additionally,an a priori information-assisted deep unfolded channel estimation neural network is modeled.Simulation results demonstrate that the proposed channel estimation method outperforms existing compressive sensing and deep learning-based methods,exhibiting significantly reduced channel estimation errors.