Optimum ber in non-orthogonal multiple access for 5G wireless technology

English Abstract: Non-Orthogonal Multiple Access (NOMA) is a promising technology with the potential to provide high spectral efficiency, reduced latency, increased reliability and robust connectivity. This focuses primarily on achieving optimal bit error rate (BER) performance. It uses recurrent neural networks (RNN) to efficiently decode messages for multiple users by treating each time step as a separate subcarrier. Additionally, a deep neural network (DNN) is used to ensure that open channel prediction is unnecessary by using its ability to automatically learn patterns in the data. This is advantageous in tasks such as decoding messages in traditional NOMA systems, simplifying the receiver's operation and reducing error propagation effects. Through careful simulations, the bit error rate (BER) of the system is evaluated at varying signal-to-noise ratios in added white Gaussian noise (AWGN) and Rayleigh dispersion systems. The results show that DNN exhibits superior BER performance compared to least squares (LS), maximum likelihood (ML), and minimum mean square error (MMSE). Using a QPSK demodulation scheme with a wide SNR range, DNN exhibits superior BER performance compared to LS, ML, and MMSE for all symbol combinations for two users.