CNN-Based Channel Estimation for Extreme Scenarios in 6G and Beyond

Channel estimation plays a critical role in wireless communication, especially under extreme scenarios that pose significant challenges to reliable communication. These challenges are expected to be more severe in 6G and beyond due to the adoption of higher frequencies (millimeter-wave and terahertz bands) and the integration of high-speed terrestrial and non-terrestrial networks for ubiquitous connectivity. Conventional channel estimation techniques, such as the Least Squares (LS) and Minimum Mean Squared Error (MMSE) estimators struggle under these conditions due to their reliance on linear models and sensitivity to noise. This research investigates the use of a Convolutional Neural Network (CNN) for channel estimation in extreme scenarios. The proposed CNN architecture captures the spatial and temporal features, as well as the nonlinear patterns in the time-frequency resource grid of wireless channels, enabling robust and efficient channel estimation. Performance comparisons between the CNN-based and conventional channel estimation techniques were conducted under varying Doppler shift, delay spread, and signal-to-noise ratio (SNR) conditions. The results demonstrate that the CNN-based channel estimator significantly outperforms conventional methods, maintaining a low mean squared error (MSE) even under severe conditions. These findings highlight CNN-based channel estimation as a robust and adaptable solution for next-generation networks.