Downlink Video Communications for Enhanced Hybrid NOMA using Deep Learning Model

Abstract

Eventually, the wireless system is one of the most prominent and effectively overloaded with enormous amounts of data information due to the rising number of users. To effectively handle widespread congestion, the current iteration of multiple access (MA) will inevitably fail. Recently, the process of non-orthogonal multiple access (NOMA) has been brought to light as a promising one for 5G and beyond, since it can improve spectrum and power efficiency for a large number of users simultaneously. In this research, a downlink Hybrid NOMA (H-NOMA) is recommended that combines the advantages of TDMA and NOMA to enhance the robustness of the system and offer two resource management techniques to maximize both outage capacity and ergodic size in a 5G URLLC scenario Ultra-Reliable Low Latency Communications (URLLC), and we conclude by proposing a deep-learning-based network, ResNet-50, to cut down on the complexity and latency of 5G URLLC. The suggested systems re-assign subcarrier in a way that maximizes outage capacity rather than ergodic capacity alone, and they select the candidate user for gaining subcarrier in a novel approach. The proposed model relies heavily on NOMA deep residual video identification to guarantee precise user categorization and widespread interaction. Using the obtained data, we examine how the suggested method stacks up against the conventional NOMA in terms of BER. When compared to other methods, the proposed one achieves a 93% (bit/s/Hz) spectrum efficiency, a 92% (bit/Hz/Joule) energy efficiency, and a 91% (bps/Hz) attainable data rate.