Enhancing GF-NOMA Spectral Efficiency Under Imperfections Using Deep Reinforcement Learning | IEEE Journals & Magazine | IEEE Xplore