Channel estimation is a challenging problem for realizing efficient ambient backscatter communication (AmBC) systems. In this letter, channel estimation in AmBC is modeled as a denoising problem and a convolutional neural network-based deep residual learning denoiser (CRLD) is developed to directly recover the channel coefficients from the received noisy pilot signals. To simultaneously exploit the spatial and temporal features of the pilot signals, a novel three-dimension (3D) denoising block is specifically designed to facilitate denoising in CRLD. In addition, we provide theoretical analysis to characterize the properties of the proposed CRLD. Simulation results demonstrate that the performance of the proposed method approaches the performance of the optimal minimum mean square error (MMSE) estimator with perfect statistical channel correlation matrix.

中文翻译：

---

环境反向散射通信中的深度残差学习辅助信道估计


信道估计是实现高效环境反向散射通信（AmBC）系统的一个具有挑战性的问题。在这封信中，AmBC 中的信道估计被建模为降噪问题，并开发了基于卷积神经网络的深度残差学习降噪器 (CRLD)，以直接从接收到的噪声导频信号中恢复信道系数。为了同时利用导频信号的空间和时间特征，专门设计了一种新颖的三维 (3D) 去噪模块来促进 CRLD 中的去噪。此外，我们还提供理论分析来表征所提出的 CRLD 的特性。仿真结果表明，该方法的性能接近具有完美统计信道相关矩阵的最优最小均方误差（MMSE）估计器的性能。