Deep learning-based receiver for 5G communication system under doubly selective fading channel

Abstract

With the increasing research on the fifth generation (5G) communication systems, especially in doubly selective fading channels, receiver designs based on deep learning have attracted widespread attention. This thesis proposes a receiver design utilizing deep learning, combining Convolutional Neural Networks (CNN) for spatiotemporal feature extraction and Recurrent Neural Networks (RNN) for capturing temporal dependencies and exploiting channel dynamics. Through joint optimization and parameter training, the receiver aims to improve the bit error rate (BER) and detection accuracy. Extensive simulations are conducted in Orthogonal Frequency Division multiplexing (OFDM) systems to evaluate the performance of the proposed receiver in comparison to traditional methods. The results indicate that deep learning-based receivers demonstrate excellent reliability and performance, providing an effective solution to enhance communication system performance in time and frequency-selective fading environments.