A Review of Modern Channel Estimation Methods: Deep Learning Approaches

Recently, the utilization of wireless communication systems and the number of users has increased. This leads to the usage of wider bandwidth and higher frequencies, which causes selective fading phenomena. To compensate for these impairments, Channel Estimation (CE) is an imperative requirement in a wireless communication system. Conventional channel estimation methods became more complex, especially with an increase in channel parameters. Deep Learning (DL) has demonstrated significant potential in various areas and has paid attention to wireless channel estimation. Deep learning improves channel estimation accuracy in diverse transmission and propagation environments, reducing computational overhead. This paper presents the basic concept of channel estimation and an extensive overview of various ongoing research in deep learning approaches-based channel estimation, which can be categorized as implicit and explicit channel estimators. With an accurate deep learning algorithm, efficient channel estimation under diverse physical and communication scenarios can be obtained that outperforms conventional channel estimators. DL is a promising approach for CE in wireless communications with complex channel distortion and interferences. The survey shows that the performance of DL algorithms depends on the quality, size, and inclusivity of the utilized dataset, while the time response to the variation of the channel parameter using implicit CE outperformed the explicit CE types.