Deep Learning Oriented Channel Estimation for Interference Reduction for 5G

Abstract

The increasing demand for high-speed data services, such as mobile gaming, Augmented/Virtual Reality (AR/VR) applications, vehicular communications, Internet of Everything (IoE), and haptic internet, results in high user densification in 5G and beyond networks. Moreover, the ultra-dense user scenarios raise the challenge of increased interference due to the highly shared spatial resources and unknown Channel State Information (CSI). Therefore, the optimal channel estimation helps in interference cancellation; however, the conventional channel estimation techniques are imperfect. On the other hand, the Deep Learning (DL) approach confers the potential solution for the channel estimation. In this paper, we implement the Convolutional Neural Network (CNN) dL architecture for channel estimation over the range of values of SNR for Single Input Single Output OFDM network. The proposed DL-CNN approach demonstrates a 94.30% reduction in Mean Square Error (MSE) compared to the traditional interpolation method-based channel estimation at different values of SNR considering the dense scenario.