A Novel CNN-Based Channel Estimation Algorithm in the Presence of Phase Noise and CFO

In this letter, we propose a convolutional neural network (CNN)-based algorithm for channel estimation in the presence of phase noise and carrier frequency offset (CFO) in fifth-generation (5G) and beyond systems. The migration of these networks to high-frequency bands, such as millimeter waves, presents significant challenges due to the unsatisfactory performance of local oscillators, resulting in hardware impairments like phase noise and CFO. Additionally, in these scenarios, the presence of intercarrier interference (ICI) imposed by these issues becomes more pronounced, posing limitations on channel estimation. To address this issue, we propose a new algorithm that simultaneously considers the imposed ICI regarding phase noise and CFO when estimating the channel gain based on received pilot subcarriers. The quality of the CNN's predictions for channel estimation is assessed using mean square error, which showed a considerable improvement in end-to-end system performance, measured by bit error rate. Our proposed approach is compared to the least squares method, the optimum least minimum mean-square error estimator, and the typical deep learning (DL) algorithm with two state-of-the-art CNN architectures. To the best of our knowledge, this is the first work that addresses channel estimation in the presence of phase noise and CFO using a CNN-based algorithm.