Enhancing spectral efficiency using a new MIMO WPT-NOMA system based on wavelet packet transform and convolutional complex neural network

Abstract

A hybrid combination of Non-Orthogonal Multiple Access (NOMA) and Orthogonal Multiple Access (OMA) technologies solves various challenges beyond the fifth generation (5G) networks. These challenges include huge connectivity, low latency, and high dependability. Accordingly, this paper proposes a new strategy for NOMA Wavelet Packet Transform (WPT-NOMA) and a new Sequential Interference Cancellation (SIC) receiver based on a Complex valued Convolutional Neural Network (CVNN). In the proposed model it is assumed that the Channel State Information (CSI) is perfectly estimated by the Base Station (BS) and users. It is utilized to enhance the Spectral Efficiency (SE) and thus improve system performance. The WPT-NOMA embeds the signals of different users in one signal. Every two users are paired via a constant power allocation and then combined by WPT. WPT-NOMA has great advantages, where a low level of WPT is needed in comparison to other algorithms. Also, the proposed receiver for the WPT-NOMA system uses only one CVNN-SIC to retrieve data. Accordingly, the proposed system outperforms the conventional NOMA and other algorithms in terms of Bit Error Rate (BER), SE, Energy Efficiency (EE), and Outage Probability (OP). A CNN and a CVNN for SIC receivers are trained on simulated data to improve the accuracy of detecting signals at different Signal-to-Noise Ratios (SNRs).