IoT-Inspired Cooperative Spectrum Sharing With Energy Harvesting in UAV-Assisted NOMA Networks: Deep Learning Assessment

Abstract

Energy and spectral efficiency of Internet of Things (IoT) networks can be improved by integrating energy harvesting, cognitive radio, and non-orthogonal multiple access (NOMA) techniques, while unmanned aerial vehicles (UAVs), on the other hand, are a quick and adaptable entity for improving the coverage performance. In this article, we assess the performance of a UAV-assisted overlay cognitive NOMA (OC-NOMA) system by employing an energy harvesting based IoT-inspired cooperative spectrum sharing transmission (I-CSST) scheme. Herein, an energy-constrained UAV-borne secondary node harvests radio-frequency energy from the primary source and uses it to send both its own information signal and the primary information signal using the NOMA approach. We consider the impact of the imperfect successive interference cancellation in NOMA and the distortion noises caused by hardware impairments (HIs) in signal processing, which are unavoidable in real-world systems. We obtain the complicated expressions of outage probability (OP) for primary and secondary IoT networks using I-CSST scheme under heterogeneous Rician and Nakagami-m fading channels. We continue to investigate asymptotic analysis for OP in order to gain insightful knowledge on the high signal-to-noise ratio (SNR) slope and practicable diversity order. We also assess the system throughput and energy efficiency for the considered OC-NOMA system. Our results demonstrate the benefits of the suggested I-CSST scheme over the benchmark primary direct transmission and orthogonal multiple access schemes. We create a deep neural network (DNN) architecture for real-time OP prediction in order to combat the complications in model-based approaches. IEEE