Super-resolution channel estimation based deep learning in reconfigurable intelligent surface systems

The propagation environment may be configured using a reconfigurable intelligent surface (RIS). The channel estimate is a critical problem in implementing the RIS-aided communication system. A cascaded channel with large dimensions and complex statistics is used in a RIS-aided multi-user multiple-input multiple-output (MIMO) orthogonal frequency division multiplexing (OFDM) communication system. This research elucidates the application of deep learning (DL) for communication channel estimation. It employs a two-dimensional visualization to represent the time-frequency attributes of a rapidly fading communication channel. The objective is to decipher the undisclosed channel response by contrasting it with recognized values at designated "pilot points." We have introduced an extensive framework that integrates sophisticated image processing methods, including techniques like image super-resolution (SR) and image restoration (IR), to accomplish this goal. This method views the pilot data collectively as a low-quality image and calculates the channel using an SR system paired with a noise-reducing IR system.Moreover, a practical application of the proposed procedure is also detailed. According to the simulation results, the trained DL estimator outperforms the Least Square estimators in predicting the channel and identifying transmitted symbols, although the suggested SRIR estimator is more sophisticated. Furthermore, the DL estimator exhibits its efficacy with varying pilot densities and cycle prefix times. The findings show that this pipeline may be utilized effectively in channel estimation.

Text 119485.pdf - Published Version Available under License Creative Commons Attribution Share Alike. Download (777kB) Official URL or Download Paper: https://ijisae.org/index.php/IJISAE/article/view/4...

Actions (login required)

View Item