Title: Efficient 5G Joint Detector in Non-Orthogonal Multiple Access Networks Based On GRU Deep Learning Approach
Authors: Mohammed AL- Darhomi , Sanaa K. Al-Asadi
Volume: 9
Issue: 2
Pages: 36-44
Publication Date: 2025/02/28
Abstract:
Non orthogonal multiple access (NOMA) has emerged as a promising technology for wireless communication networks. However, optimizing its efficiency in the presence of imperfect signal detection remains a challenge. Deep learning (DL) algorithms offer the potential for improving signal detection and channel estimation, and this work proposes an enhanced NOMA detector using DL gate recurrent units (GRU) to address the issue of vanishing or exploding gradients commonly encountered in DL algorithms. The proposed approach surpasses competitive DL systems in terms of sample error rate (SER). Extensive testing with various parameter configurations demonstrates its superiority over enhancement LSTM by approximately 2dB and BILSTM by about 4dB. It also outperforms traditional channel estimation methods such as least squares (LS), minimum mean squared error (MMSE), and maximum likelihood (ML) at a low signal-to-noise ratio (SNR). Remarkably, even at high SNR, our approach achieves performance comparable to ML while outperforming other signal detection methods. Moreover, the proposed GRU-based model reduces the number of gates used, resulting in a simpler architecture with reduced inter-dependencies between users. This reduction in dependency can lead to improved performance and faster training times. Additionally, the model offers reduced training time compared to other DL systems, enabling efficient practical deployment.